Monthly Roundup

HighLevel Updates: October 2026 Roundup

October closed with four release days in a row. October 30 added Hosted Numbers in Labs, which brings a US local or toll free number a client already uses into HighLevel for SMS while voice stays with its current carrier, usage limits for SaaS clients on Voice AI and Conversation AI with a request and approval flow when a client needs more, and three call changes on mobile: live connection quality with a star rating afterwards, every recording on a multi-leg call instead of only the first, and Answered By showing who picked up and on which device. On October 27 more than sixty features shipped at once, led by a Conversation AI that opens conversations instead of waiting for them, a visual map of every automation in the account, an email builder that works from a prompt and HighLevel's own video conferencing. October 28 followed with workflows exposed through MCP and a REST API, an AI builder that ships branded Android apps off the client portal, Managed Agents that delegate to subagents and drive a browser, deposits, fees and add-ons for rentals plus RCS messaging in Labs. October 29 added four workflow actions, Router for branches that all run, Loop for repeating a block of actions over a list, AI Video Generation on Veo 3.1 and AI Analyze Video on Gemini, gave the AI Agent action a skills library and a web search toggle, let AI Builder test the workflows it writes, opened service booking inside the client portal and gave the calendar view a settings panel with PDF and CSV export. The rest of the month produced a WhatsApp migration deadline, a price rise called off the day before it landed, a storefront, a backend, a schema dashboard and one genuinely useful report. On October 19 HighLevel confirmed that Meta stops delivering marketing messages through the Cloud API on October 13, 2027, so every WhatsApp marketing send has to move to the Marketing Messages API, which HighLevel already supports as MM Lite. Service and utility messages are unaffected, activation takes minutes per sub-account, and existing numbers and approved templates carry over untouched. On October 12 HighLevel cancelled the voice pricing update announced on October 1, so inbound minutes in 19 countries and outbound minutes to 21 destinations keep their current rates. The messaging and phone number rental increases from the September 21 notice are separate and still take effect on October 13. The same day, websites and funnels gained an SEO and AEO tab with a schema dashboard that scores coverage across the site, flags pages whose schema has gone stale and generates schema with AI in bulk, and the Meta Pixel started firing InitiateCheckout when a buyer reaches checkout instead of when they pay, which exposes real checkout drop-off for the first time. On October 9 AI Studio projects gained Cloud, a private backend created in about a minute with a database, user accounts and file storage, priced by the hour and free at the smallest database size. On October 8 AI Studio apps gained payments, selling catalog products through HighLevel's hosted checkout with one-click post-purchase upsells, and every published AI Studio site started reporting its own page views, opt-ins and form submissions with no tracking code. On October 6 the Lead Source Report gained a monetary value for every status and a drill-down from any count to the opportunities behind it, phone numbers started showing their real A2P carrier registration status instead of only that they had been submitted, and recurring appointments stopped sending a separate confirmation for every occurrence. A2P rejection guidance now matches how the campaign was registered, and Social Planner can publish Community posts through the public API. On October 10 the custom field types released in September gained snapshot and public API support, so field configurations replicate to new sub-accounts and integrations can write to them. On October 14 workflows gained an Add Internal Comment action, which posts internal context straight into a conversation and can @mention the user who has to act on it, AI Studio apps gained a native HighLevel chat widget that the agent builds into the app and reports on, and agencies gained bulk feature control, enabling or disabling features across many sub-accounts in one pass. On October 15 the WhatsApp template builder was rebuilt with a live phone preview and media headers that accept a custom field, so one approved template sends every contact their own invoice, ticket or personalized image, and a new limited time offer template type puts a countdown WhatsApp renders itself inside the message with the expiry set at send rather than at build. The same day AI Studio projects gained two-way GitHub sync with a private repository and secret scanning, the AI Studio checkout gained field control, a two-step lead-first option and order details after payment, and gift card confirmations, emails and SMS started translating from the selected locale. Carried in from the last week of September: Ask AI on mobile now holds a spoken conversation, invoices have an AI layout builder, general dashboard widgets accept permanent filter conditions, Company and Custom Object Smart Lists travel inside snapshots, Ad Manager controls Meta campaign budgets, Community email unsubscribes are scoped to one community and one notification type, and Mexico local numbers have lost SMS entirely. On October 16 number pool call tracking started labeling calls by ad platform, so paid Facebook calls read Facebook Paid instead of Facebook or Direct, TikTok, LinkedIn, Microsoft Bing and in-app Google clicks carry their own labels, ChatGPT referrals are counted as a source of their own, and a number pool can ring up to six selected users at once on whichever device each of them prefers. On October 22 the HubSpot importer gained date filters on its preview step, so a migration brings the active records and leaves the rest, and service and rental bookings created inside the app can carry a completed booking form. On October 25 Workflows gained a Social Planner Posts trigger, firing on Approval Requested, Approved, Approval Rejected, Scheduled, Published and Failed, with filters for category, platform, post type, connected social account and post tag, and support for Integrations-connected accounts still pending. On October 26 a reputation and social batch landed. Reviews AI agents gained brand voice controls, quality guardrails, a hold for sensitive reviews and a live response preview, agencies can set Reviews AI to Off, Suggestive or Auto across many sub-accounts at once, and Location Dashboards gained 16 reputation widgets. Social Planner gained Social Intelligence in limited beta, a password-protected invite that lets clients connect their own social accounts with no sub-account access, and comment sync that brings Facebook, Instagram and TikTok comments into the Conversations team inbox as contacts you can answer publicly or by DM. Conversations gained keyword search across the inbox and inside a single thread, plus six new filters covering channel account, last call status, contact type and opportunity pipeline, stage and status. The same day, WhatsApp gained five releases of its own: carousel templates holding up to ten swipeable cards for the price of one marketing message, forms that open inside the chat as WhatsApp Flows, call recording with transcripts and a consent announcement, a library of 52 pre-built templates in six languages, and appointment booking flows in 13 languages. Every WordPress site gained Antivirus with ten checks and ten one-click hardening toggles, a rebuilt per-site menu and a sites table carrying threats, uptime, backups and performance, and blog posts gained reader comments with moderation. On October 27 profile and business cards started generating themselves behind a new QR icon in the top bar, with Exchange Contact writing a scan straight into the CRM, and QR analytics added device, browser and city-level geography. October 27 was the release day itself. Conversation AI gained outbound triggers, AI Campaigns for bulk reactivation, rich elements on WhatsApp, Messenger, Instagram and Live Chat, MCP server access to apps like Google Calendar, Sheets, Docs, Canva and Zapier, and a model menu that now includes Claude and Gemini alongside GPT. Workflow Maps drew every workflow, folder and connected asset on one canvas, with contact journeys, error detection and an assistant, while workflow testing became step by step and an A/B Test action arrived. Email Marketing was rebuilt with an AI editor, AI-built sequences, a pre-send spam check, real inbox previews and domain warm-up inside LC Email. The CRM made domain-based company creation and association labels generally available, with associations in Smart Lists, an Association Created trigger and calculated fields in private beta. Forms gained an Integrate tab, product-level conditional logic, automatic taxes and the ability to create an opportunity with no contact attached. Ad Manager added TikTok and two-way Meta sync, Meeting Room arrived with breakout rooms, Voice Intelligence started writing call notes in Labs, and Affiliate Manager gained 32 public API endpoints, audit logs and failed payout recovery. On October 28 Conversations panels became resizable and Ask AI Computer reached the mobile app.

30-day free trial through our link (the public site offers 14) · No contracts · Cancel anytime

Published 2026-10-30 · Updated 2026-11-16

October ended with four release days in a row, and the last one was about getting paid and knowing what happened. Hosted Numbers entered Labs, so the number a client already advertises can start sending SMS in HighLevel without porting the voice service with it. SaaS usage limits arrived for Voice AI and Conversation AI, with a cap per client, a billing page that shows how much is left and a request you approve or reject when a client wants more. The mobile app gained live call quality with a four-bar indicator and a post-call rating, separate playback and transcripts for every recording on a transferred call, and an Answered By line naming the person and the device that took the call. On October 27 more than sixty features shipped in a single day, and the ones that matter point the same way: the AI stopped waiting to be asked. October 29 then handed the same idea to the workflow builder. Router sends a contact down every branch it matches instead of stopping at the first. Loop repeats a block of actions once per item in a list or once per number in a range. Two new premium AI actions generate a short video from a prompt and answer questions about one. The AI Agent action gained a skills library it loads only when a run matches, plus a web search toggle billed per search at 1 cent on OpenAI and Claude models and 1.4 cents on Gemini. AI Builder started testing the workflows it writes inside the same chat, the client portal took over service booking, and the calendar view got a settings panel that remembers your timezone, grid interval and colors. Before that the month had already delivered one WhatsApp migration deadline, one withdrawn price rise, a storefront, a backend, a GitHub sync, a rebuilt WhatsApp builder, a schema dashboard, a social listening workspace, a searchable inbox and one better report. The newest item is the one that needs a decision. On October 19 HighLevel confirmed Meta's cutoff: from October 13, 2027 marketing messages stop being delivered through the Cloud API, and every WhatsApp marketing send has to run on the Marketing Messages API, which HighLevel already supports as MM Lite. Service and utility traffic is unaffected. The deadline is twelve months out, but Cloud API is frozen for marketing, so the cost of waiting is paid now, in delivery optimization, funnel tracking past read receipts and formats like animated GIF headers that exist only on MM Lite. The other headline is a reversal. On October 12, one day before it was due, HighLevel cancelled the voice pricing update it had announced on October 1, after the underlying providers confirmed the change was off. Voice rates stay as they are. The messaging and phone number rental increases from September 21 are a separate notice and still land on October 13, so the audit is narrower than it first looked. The same day, every website and funnel gained an SEO and AEO tab where schema is managed for the whole site rather than page by page, with a coverage score, a Might be outdated flag for pages republished since their schema was written, and bulk AI generation you review before publishing. Also on October 12 the Meta Pixel moved InitiateCheckout to the moment a buyer reaches checkout instead of the moment they submit payment, so the step between starting checkout and entering card details finally shows drop-off, and InitiateCheckout counts will rise. AI Studio took two of the earlier days. On October 8 apps built there could sell products from your catalog through HighLevel's hosted checkout, with PayPal express, coupon codes, subscriptions and one-click upsells on a saved card, and every published site started reporting page views, opt-ins and every form submission without a tracking code. On October 9 each project gained Cloud, a private backend with a database, user accounts and file storage, switched on from More, then Cloud, and running about a minute later. On October 6 the Lead Source Report started showing monetary value for Open, Won, Lost and Abandoned, and any count above zero now opens the opportunities behind it. The same day A2P registration status went live on the phone numbers page, which ends the habit of sending from an unregistered number and collecting error 30034, and recurring appointment notifications were rebuilt so a series sends one confirmation and custom recurrences send anything at all. On October 10 the custom field types released in September, URL, Date and Time, Rich text, Users and image options in Radio and Checkbox fields, gained snapshot and public API support, which takes field setup out of manual sub-account onboarding and lets an integration create and update them. On October 14 workflows gained an Add Internal Comment action, so a workflow can post internal context into a conversation and @mention the user who has to act on it, which takes handover notes off the person who happened to pick the thread up. The same day AI Studio apps gained a chat widget that the agent builds into the app in your site's own colors, with its own lead table and chat analytics, and it runs on React projects as well as TanStack Start, so no migration is needed. Also on October 14, features can be enabled or disabled across many sub-accounts at once from the Sub-Accounts page, with every request logged in Bulk Actions History. October 15 was the busiest day of the month. The WhatsApp template builder was rebuilt into one layout with a live phone preview, and media headers now take a custom field instead of one hard-coded URL, so a single approved template can send every contact their own invoice, ticket or personalized image and a different asset no longer means a different template waiting on Meta. A limited time offer template type arrived alongside it, with a countdown WhatsApp draws inside the chat, a discount code that copies in one tap and an expiry chosen at send time, so one approved template runs campaign after campaign. AI Studio took the rest of the day. Projects can connect a GitHub account under More, then Git, which creates a private repo, pushes every saved version and pulls back commits made outside, so the code can be edited in any editor or coding agent and come back with a fresh preview, with pushes scanned for keys and a new Manage Git and download code permission on by default. The AI Studio checkout became configurable too: show, hide or require phone and each billing address field, split it into a two-step checkout that saves the lead before taking the payment, and show order details after payment or send buyers to your own thank-you page. Watch one thing there, because two-step leads fire Contact Created rather than Order Form Submission. Gift card purchase confirmations, recipient emails and SMS now translate from the selected locale, previews included. The rest of this page covers the releases from the final week of September that you are actually using in October: two-way voice Ask AI on the mobile app, an AI layout builder for invoices, dashboard widgets with filter conditions that persist, snapshot support for Company and Custom Object Smart Lists, Meta campaign budget control in Ad Manager and scoped unsubscribes for Community email. October 16 closed the week on call tracking. Number pool calls now name the ad platform that produced them, which ends the habit of reading a Facebook or Direct label and guessing whether the spend or the page did the work, and a pool can ring up to six named users at once on web, mobile, a desk phone or their own number. October 22 added two smaller fixes that each remove a cleanup job. The HubSpot importer now filters Contacts, Deals, Companies and Tasks by created or last modified date before anything crosses over, with the record count updating as the filters change, and service and rental bookings created from the in-app appointment modal can carry a booking form that staff complete on the customer's behalf, prefilled from the contact and stripped of CAPTCHA and payment elements. October 25 added a Social Planner Posts trigger in Workflows. It fires on Approval Requested, Approved, Approval Rejected, Scheduled, Published and Failed, filters by category, platform, post type, connected account and tag, and finally gives a failed post somewhere to raise its hand. Only accounts connected through Social Planner work with it so far. October 26 brought eight releases across reputation, social and the inbox. Reviews AI agents became configurable: a response style of we or I, the business name, tone traits, a length limit and a sign-off, plus guardrails that remember recent replies, avoid repeating an opening line and refuse refund promises, guarantees, admissions of fault and medical or financial advice. Sensitive reviews can be held for a human, and a preview shows how the agent answers a positive, a neutral and a negative review before the configuration is saved. Agencies set the mode in bulk from AI Suite, then AI Setup, and Location Dashboards gained 16 reputation widgets, which is the first time review work has had a client-facing number attached to it. Social Planner gained Social Intelligence in limited beta, where brand mentions, industry trends and share of conversation against up to three competitors sit in one workspace and the old Social Listening lives on under a Trends tab, and a Social Connect invite that hands a client a password-protected link so they connect their own accounts without touching your sub-account. Comments on Social Planner posts now arrive in the Conversations team inbox from Facebook, Instagram and TikTok, every commenter becomes a contact, and a reply can go public under the comment or private as a DM. Conversations also gained the thing it was missing, keyword search, across every conversation and inside a single thread, plus filters for channel account, last call status, contact type and opportunity pipeline, stage and status. October 26 also rebuilt WhatsApp around what a message can carry. A carousel template holds 2 to 10 cards, each with its own image or video and up to two buttons, and WhatsApp bills the whole thing as one marketing message and counts it once against frequency limits, so a tap on card three now tells you which product landed. Forms open inside the conversation as WhatsApp Flows, which removes the redirect that quietly loses leads, and answers land in Forms, then Submissions, attached to the contact. WhatsApp calls can be recorded and transcribed with a consent announcement played first in the language you choose. A template library of 52 pre-built messages covers the usual industries in six languages, and appointment booking flows now render in 13. On the hosting side, every WordPress site gained Antivirus, which runs ten checks, explains each problem in a line and fixes it on confirmation with the last backup date shown, plus ten one-click hardening toggles and a threats count on the sites list. Each site now opens to one menu with tools grouped by job, and the sites table carries status, threats, updates, uptime, backups, performance and storage. Blog posts gained reader comments, with sign-in by one-time code, a Moderator badge on your replies and a Comments tab for approval, though comments are per blog site rather than per post and do not reach Conversations yet. October 27 added the smallest and most immediately usable item of the month: a QR icon in the top bar holding a Profile Card built from your user details and a Business Card built from the sub-account's, both already filled in, each with its own link, QR code and scan count. Save Contact is on by default and Exchange Contact, once enabled, turns a scan into a contact in the CRM with the card recorded as its origin. QR analytics gained device, browser and country, state and city reporting at the same time. The October 27 release day is covered section by section below, starting with Conversation AI, which can now open a conversation on a trigger, run a reactivation campaign across a whole audience, call your other apps through MCP servers and run on Claude or Gemini instead of GPT. Workflow Maps follows, then the Email Marketing rebuild, the CRM association releases, the form builder changes, Ad Manager's TikTok and Meta sync, Meeting Room and Voice Intelligence, and the Affiliate Manager batch. October 28 then matched it. Workflows became readable and writable from outside the platform, through the LeadConnector MCP server and a REST API, both in private beta, so an assistant or your own code can find a workflow, read its definition, build one from a description, publish it and report which steps failed and for whom. The Client Portal turned into a builder: describe the experience and AI generates the pages, or generate a branded Android app with sign-in, join codes, camera access and an installable APK. Managed Agents moved into one workspace, learned to delegate to subagents running in parallel, gained a Computer that browses sites, writes files and runs Python, and can be created by importing an n8n workflow export. Rentals can take a deposit and charge the balance later, carry mandatory fees and optional add-ons, service appointments can sell retail products off the catalog, and appointment statuses can be named in your own words. RCS registration moved inside HighLevel through Labs, A2P 10DLC got one agency dashboard covering every sub-account, the website builder gained one-click QR codes, Conversations panels became resizable and Ask AI Computer arrived on the mobile app.

October is closed and this page is the final record of the month, October 1 to October 31. September is documented in full in the September roundup.

Hosted Numbers brings an existing number in for SMS

October 30 opened Hosted Numbers in Labs, so a US local or toll free number a business already uses can send and receive SMS in HighLevel while voice stays with its current carrier. Bringing a recognized number into the platform used to mean porting it, which moves the voice service too.

  • Voice is untouched. Calls keep running through the number's existing carrier during the process and after it. Only messaging moves
  • The setup is guided. Check whether the number is eligible, verify ownership through an automated call, pick or add the service address, then send a letter of authorization for electronic signature. The request then tracks through verification, authorization, processing and final checks, and hosted and in-progress numbers show up in both Hosted Numbers and Phone Numbers
  • Hosted is not the same as SMS ready. Hosted means the hosting process finished. SMS ready means location identity verification where it applies, location-level A2P registration and messaging sender setup are all complete. If the sub-account already has an approved A2P registration, HighLevel reuses it, so there is no second registration and no campaign to pick per number
  • Toll free numbers need toll free verification on top of the hosting request
  • Agency owners and admins start it. Sub-account users can find Hosted Numbers and watch requests for their own location but cannot initiate one. Readiness steps stay with whoever already has permission for identity and A2P work
  • The Labs release is narrow. Eligible US local numbers, SMS hosting, one number per request. Enable Hosted Numbers in Labs, then Settings, Phone System, Manage Numbers, Add Number, Host a Number

This is the release for the client who will not give up the number painted on the van. Porting has always been the answer and always the objection, because the voice line goes with it. Hosting separates the two. Watch the readiness split in practice, because a hosted number that is not SMS ready will not send. The A2P work still has to happen, it just does not have to happen twice.

SaaS clients can be capped, and billed when they ask for more

Also on October 30, usage limits arrived for SaaS clients, starting with Voice AI and Conversation AI. Each client gets a limit, and you decide what happens when they reach it, bill the extra usage or stop the action.

  • Clients see their own usage. The billing page shows consumption against the limit with color cues as it gets close, plus the date usage resets
  • You see who is near the ceiling. Manage Client lists every client with a limit, filtered by product or by how much of the limit is gone
  • Limits can be set at onboarding. Adding a usage limit is an optional step while setting up a new client with rebilling, so it exists from day one rather than after the first surprise
  • Clients can request more. A client who cannot upgrade their plan can ask for a higher limit from the billing page or the limit prompt. The request appears in Manage Client, on the agency dashboard and in the SaaS Configurator, and you set a number you are comfortable with and approve it, or reject it with a reason, without leaving the page
  • Both sides get notified. You get an in-app and email notification when a client asks. Clients hear when their limit changes and when a request is rejected, with your reason attached
  • It is a Pro agency Labs beta. Voice AI and Conversation AI only. Only sub-account admins can view usage and request more, and the request option appears only for clients who cannot upgrade their plan. Approving a request changes the limit and nothing else, so the behavior at the limit stays as you set it. The Enable Rebilling flow and the bulk action do not carry usage limits yet

This is the missing half of rebilling. Reselling AI usage has meant either absorbing a client's heavy month or having the conversation after something stopped working. A cap plus a request flow turns that into a yes or no you can price. HighLevel says alerts at 50, 80 and 100 percent are coming, along with limits for the phone system, email and automations, which is where this gets genuinely useful, because Voice AI and Conversation AI are not usually the only line that moves a margin.

Mobile calls show quality, every recording and who answered

October 30 brought three call changes to the mobile app, all on version 4.27 and later across HighLevel, LeadConnector and self-serve white-label builds, on iOS and Android.

  • Signal bars during the call. A four-bar indicator appears as soon as the call connects and turns green, orange or red as the connection changes. Tapping it opens Connection quality, with a plain-language summary and separate Good, Weak or Poor ratings for Network, Device and Audio. Device covers connection type, Wi-Fi strength, phone temperature, battery saver and low data mode. Audio covers the mic, incoming audio and where sound is playing
  • One tip, not a diagnostic dump. When the connection slips the app shows a single suggestion that matches the situation, move closer to the router, switch to Wi-Fi, turn off battery saver. Get help holds short answers for they cannot hear me, I cannot hear them and audio keeps breaking up
  • A star rating after the call. Four or five stars sends immediately. Three or fewer asks what went wrong, with choices for could not hear them, they could not hear me, choppy or robotic audio, call dropped and something else with an optional note. The sheet opens on its own at most once a week, and Rate this call on the call end screen covers any other call
  • Every recording, not just the first. A call with more than one recording now shows each one separately, labeled 1/2 and 2/2, each playing on its own with its own transcript. Transferred and multi-leg calls stop losing the leg that mattered. A transcript appears only where the backend has one
  • Answered By in call details. Long-press a call message in Conversations, then Details, and the call shows who picked it up and the device they used, web app, mobile app, phone number or SIP. Regular, IVR and WhatsApp calls are covered, numbers are masked when a call was answered on a phone number, and no icon shows when the device is unknown

The last one is the useful one. With ring groups and forwarding the assigned user and the person who answered are routinely different, and on mobile there was no way to tell which. Now coaching and follow-up have a name attached. The soft reconnect tone also replaces the old vibration and banner on phone and WhatsApp calls, call ratings for WhatsApp calls are next, and the same call context is due on call logs so a call can be understood without opening Conversations.

Workflows gained Router, Loop and two video actions

October 29 added four workflow actions, and two of them change the shape of a workflow rather than what it can send. Router and Loop cover branching and repetition, which previously meant a second workflow or an outside tool.

  • Router runs every branch that matches. If/Else sends a contact into the first matching branch and skips the rest. Router checks all of them and runs each match in turn, left to right. Branches come in three types: Custom rules with conditions set under Only continue if, Always run with no conditions and Fallback for contacts that matched nothing else. You set the order by dragging branches in the Branches list, and the test output lists every branch condition as passed or not, with the reason and the value the contact actually had
  • Loop repeats a set of actions per item. Loop Over takes a list of items or a range of numbers. The list can be an array from an earlier step such as an Inbound Webhook, and inside the loop Custom values, then Loop Item, maps the current item's fields into any action. Execution is sequential by default with a Parallel toggle, and the Test tab shows every iteration and every action inside it. One webhook carrying three services can now create three opportunities, three tasks and three notes
  • AI Video Generation writes a short video from a prompt. Veo 3.1 Quality or Veo 3.1 Fast, 720p or 1080p, landscape or portrait, and a duration of 4, 6 or 8 seconds, with a Generate Audio toggle for ambience or background music. Up to three JPEG or PNG reference images set subject and style, four templates cover product mockups, social posts, promo flyers and lifestyle, and Enhance Prompt expands a one-line brief. The video lands in the Media Library and the action returns its URL for the steps that follow
  • AI Analyze Video answers a prompt about a video. Gemini 3.8 Flash or 3.5 Flash Lite watches the picture and listens to the audio, then returns text. It takes public YouTube, youtu.be and Shorts links or a direct MP4, MOV, WebM, MPEG, FLV, WMV or 3GP file, and the URL can come from a previous step, so the generation action can feed the analysis action in the same workflow. Ask for a summary, a transcript, a timestamped list or a yes or no check
  • Both video actions are premium. Each execution carries an additional charge driven by the model, the length of the video and, for analysis, the length of the answer. The video URL passed to AI Analyze Video has to be public, so a page link or a private file will not work

These join the AI actions already in workflows, and the pricing pattern holds: the expensive steps are billed per execution rather than bundled into the plan.

The AI Agent action learned skills and web search

The same day, the AI Agent action gained a Skills library of reusable procedures and a Built-in tools section whose first entry is web search. AI Builder also started testing the workflows it writes, in the chat window where it wrote them.

  • Skills are procedures the agent loads on demand. A skill has a name, a line answering when the agent should use it and a body of instructions. On every run the agent reads only the name and that when-to-use line, and the full instructions load only when the run matches, so a long library costs nothing on an unrelated run. When a run matches more than one skill they load together, which is how one outreach email can run Cold-email and Humanizer at the same time
  • Three ways to add one. Create with AI asks what the skill should do, when it should fire and what it must never do, then drafts it for review with nothing saved until you confirm. Write skill instructions gives you the three fields directly, with Enhance Prompt for the wording. Add existing skill searches the library and attaches one you already wrote
  • Four skills ship with every account. Humanizer strips the known signs of AI writing, Write-well keeps a message clear and direct, Cold-email writes first touch and follow-up messages to leads who have not replied and Sms-writer keeps texts short. They are tagged Built-in and can be edited or deleted like any other skill
  • Web search is a toggle in Built-in tools. Turn it on and the agent looks things up mid-run, but only when the task needs facts it does not already have. Paste a URL into the Instructions and it reads that page when the run calls for it, which is the difference between an agent quoting a rebate program correctly and one guessing at it. Billing is per search: 1 cent on OpenAI and Claude models, 1.4 cents on Gemini
  • AI Builder now tests what it builds. Every build or edit reply ends with a Test workflow button. Choose Enroll a contact or Test using events, start the run, and the canvas marks each step Running, Success, Action skipped or End while the chat lists the run step by step and then writes a summary of what ran, what was skipped and where data was missing. Clicking any step on the canvas opens its Input and Output. It needs The New Test Workflow switched on in Labs
  • A test run is a real run. Tags get added, notes get written and messages get sent, so test against your own contact record rather than a customer's

Image and video generation are both listed as coming to Built-in tools, which would make the agent action a superset of the separate AI actions rather than a sibling of them.

The client portal took bookings, and the calendar view got settings

Two October 29 releases cover the client-facing and the staff-facing side of the same calendar. Clients can book services inside the portal, and the internal calendar view finally has a settings panel that remembers what you chose.

  • Service booking in the client portal. Clients browse available services, pick a date and time and book without leaving the portal, with their details prefilled and single or multiple services depending on how the booking page is set up. Rescheduling and cancellation follow the calendar's own rules, and meetings and service bookings now sit together under Appointments, split into Upcoming and Past. Visibility is per booking page, under Calendars, then Services, then Booking Pages, where Allow booking from client portal is the switch
  • One panel for calendar display settings. The three dots menu next to Manage view opens Calendar view settings, and the choices persist. View appointments in a timezone of your choice, which also applies to your appointment emails and affects nobody else. Set the day and week grid interval to 15, 30 or 60 minutes. Grey out hours outside each staff member's working schedule for that date. Hide empty rows and columns. Show buffer time wherever a slot is blocked by one
  • Timezone moved house. View in timezone has left Meetings Preferences and the Services and Rentals global settings and now lives in this panel, carrying over any value already set
  • Colors you can read at a glance. Admins get a Staff colours tab to recolor appointments per staff member, event tiles can be filled with the calendar or service color or the staff color with the other one on the left strip, and that strip is wider than it was
  • Export the view you are looking at. The same three dots menu offers Export to PDF or Export to CSV, carrying the current filters and date range
  • Nothing customer-facing changed. Availability, booking rules and the public booking page are untouched. These are display and workflow settings only

This is the third client-facing portal change of the month, after the courses and offers storefront and the CRM-connected custom objects in the app builder.

Workflows reached MCP, an API and Slack triggers

Two private betas on October 28 opened workflows to tools outside the platform, one through the LeadConnector MCP server and one through a REST API. Both cover the same ground, and both read execution data, which is the half that matters once a workflow is live.

  • Workflows in MCP. Claude, ChatGPT, Cursor and other assistants connected to the LeadConnector MCP server can search workflows by name, status, folder or date, read a full definition including triggers, steps, branches and settings, build one from a description and save it as a draft, update and publish it or set it back to draft, and delete one after a second confirmation
  • Workflows API. The same operations from your own code, plus a validate call that checks a definition and points at the exact failing step before anything is saved, and a step types endpoint that lists every trigger and action available in the sub-account with its fields. The obvious use is rolling one workflow out across many sub-accounts without rebuilding it by hand
  • Logs and statistics from either side. Who entered a workflow, what happened at each step and which steps failed with the error, plus enrollments, completions and failures per step by hour, day or week. Asking which steps failed this week replaces reading execution logs one at a time, and a Monday failure report can run from your own monitoring
  • Five Slack triggers. Workflows start on a direct message to your Slack app, a user joining or being invited to a channel, an app mention, a new message in a channel or an emoji reaction. Add trigger, search for Slack, pick the connected account
  • Workflow Maps gained onboarding. First visit explains the map and offers 2D or 3D, prompts introduce features in context, and a Help panel holds walkthroughs for tracing a contact, reviewing errors and checking asset usage, with a cheatsheet, a supported assets list and an FAQ

The two betas are the first time workflow structure has been readable and writable from outside the builder, so agencies running the same automation across many clients can finally treat a workflow as a deployable artifact rather than a thing to recreate.

Managed Agents got one workspace, subagents and a computer

Four releases on October 28 rebuilt Managed Agents in Agent Studio and widened what an agent can actually do. The n8n importer is the one that changes the maths for anyone already running automations somewhere else.

  • One workspace instead of several pages. Agents list on the left with search, folders and sorting by name, creation date or last update, the chat and test pane sits in the middle, and settings, publishing, activity and configuration sit on the right. Draft and published agents are labeled, and switching between agents no longer means navigating away
  • Subagents split the work. An agent breaks a complex task into focused pieces, hands each to a subagent holding only the tools and context that piece needs, and runs the independent ones in parallel while the main agent plans, reviews and answers once. Smaller context per subagent and lighter models on the narrow tasks also lower the execution cost
  • Computer browses, writes files and runs code. An agent can work a website, click buttons, fill forms and complete multi-step tasks, create, edit and export documents, spreadsheets and reports, and run Python scripts and terminal commands, with a persistent workspace so files carry between conversations. You can watch it work and take the browser over
  • n8n workflows import. Upload an n8n export or paste the JSON and HighLevel maps webhooks, schedules, forms and appointment events to HighLevel triggers, IF and Switch nodes to decision rules, Wait nodes to scheduled follow-ups, AI steps to the agent's own instructions and app integrations to Marketplace connections, CRM actions or API calls. Imported agents land in an Imported from n8n folder, and the mapping is shown before anything is created

The client portal became an app builder

Three releases on October 28 turned the Client Portal from somewhere to host courses into something you design, and then ship as an Android app. All three work from a prompt, and all three stay wired to the members, courses, communities, files and appointments already in the sub-account.

  • Client App Builder. Describe the business and the experience and AI generates the screens, design and functionality of a branded Android app, with a design system and brand board carrying the logo, colors and images. Sign-in, sign-up and member onboarding with join codes are built in, camera and photo access are supported, a live phone preview shows the result, and the builder produces an installable APK with build progress and errors visible
  • Vibe Coding for the web portal. The same approach for the portal itself. AI asks questions, generates working pages and takes visual edits where you select an element in the preview and describe the change. Up to 10 custom domains, light, dark or system theming in the builder, and a start either from scratch or from the portal you already run
  • Custom objects in the portal. Describe what members need to submit and AI designs the objects, fields, dropdowns and relationships, then generates pages where members view and add records such as service requests, applications or inquiries. Submissions save as custom object records in the sub-account CRM, visibility is set to a member's own records, their company's records or read-only access to all, and a record can take a payment through invoices where a provider is connected. Three objects per request, against the standing limit of 10 custom objects per sub-account
  • Discover turns the portal into a storefront. Courses, offers and communities appear in a Discover section with thumbnails, pricing, descriptions and tags, searchable and filtered by Free, language, difficulty and topic, with a Show in Discover toggle deciding what appears and checkout one click away. Discover can be set as the portal's landing page
  • Course video analytics. Views, watch time, average engagement and finish rates with comparison against the previous period, then per video unique viewers, rewatch rates, completion funnels and retention heatmaps showing where learners drop off, skip ahead or rewatch. Course AI reads the same data and recommends what to change, and the last 30 days of metrics appear inside the lesson editor

Rentals learned deposits, fees and add-ons

Five calendar releases on October 28 close the gap between taking a booking and collecting the money. Rentals can hold a reservation on a deposit, carry mandatory charges and optional extras, and service appointments can sell product off the shelf.

  • Partial payments. Set a deposit of 1 to 99 percent of the booking total in Global Settings and say when the balance is due. Checkout shows Due today and Due on the date with a tooltip explaining the automatic charge, and on the due date the balance can be charged to the card used for the deposit. A security deposit collected upfront can be taken at booking or with the final payment
  • Additional fees. Reusable mandatory charges such as cleaning, resort, service or damage waiver fees, built once under Calendar settings, then Rentals, then Additional fees, and attached to as many listings as needed. Flat or percentage pricing, flat fees charged per booking or per booking unit, tax support, per-listing overrides, and every fee shown in the booking widget and as a line item on the invoice
  • Add-ons. The optional half of the same system: a pet fee, a GPS unit, a guided tour, equipment insurance. Each carries a name, description, image, pricing, maximum quantity and tax settings, customers pick quantities in their own section of the booking widget and watch the total update, and each selection appears on the invoice as its own line
  • Retail products on service appointments. Add a product or variant from the catalog while creating or editing an appointment, set the quantity and override the unit price for that appointment without changing the catalog price. Products sit beside services in the appointment details and the Payments tab under one total, inventory drops when a product is added and returns when it is removed, and the payment status says whether more is due or a refund is owed
  • Custom appointment statuses. Name your own stages, such as Checked in, Awaiting documents or Ready for appointment, each belonging to one of six states. Pending and Active hold the time slot, Completed, No show, Cancelled and Invalid release it, and moving between statuses inside the same state only changes the label without firing notifications or calendar updates. The six existing statuses carry over as the defaults in their matching states

RCS entered Labs, and A2P got an agency dashboard

Two compliance releases on October 28, both of which remove a trip to someone else's portal.

  • RCS messaging in Labs. Agencies enable RCS per sub-account from Labs and complete the whole sender registration inside HighLevel: brand identity with business name, logo, description, contact information and website plus a preview of how the brand appears to customers, the countries where RCS will be used, and sender details covering the use case, the message types and how customers give consent. No external provider platform in the loop
  • Agency Trust Center. One dashboard in Agency view for A2P 10DLC across every sub-account, tracking Fully Approved, In Review and Rejected registrations, naming the common brand and campaign rejection reasons with a route through to the affected registrations, and comparing chat widget against self-serve registration adoption and approval rates. Brand registrations are viewed and managed in the same place, so a compliance problem surfaces before a client's sends start failing

Website QR codes and resizable conversation panels

Two small October 28 changes that each save a step.

  • QR codes in the website builder. A QR icon next to the page URL creates, downloads or customizes a code for the page being edited, the way funnels already worked, and the QR Code Builder gained a Website type where you pick a website and a page from two dropdowns. The code keeps resolving to the right page after a slug or domain change
  • Resizable panels in Conversations. Drag the dividers between the conversation list, the message thread and the contact details panel. Widths are saved per user, with minimums and maximums so the layout stays readable

Conversation AI stopped waiting to be messaged

On October 27 HighLevel released its largest Conversation AI batch of the year, and the headline is direction: the agent can now start conversations instead of only answering them. Outbound triggers, bulk campaigns, rich message formats, external tool access and a choice of model all landed the same day. Read together they turn Conversation AI from a reply bot into something that runs its own outreach.

  • Outbound agents with triggers. Under Deploy, then Outbound, an agent reaches out first when a form is submitted, a Facebook or Instagram lead ad arrives, a call is missed, a tag is added, an appointment is booked or a deal changes stage, then follows up on its own when the contact goes quiet. There is no workflow to build or maintain
  • AI Campaigns for database reactivation. Select an audience from the Contacts page and the agent writes a personalized first message per contact from that contact's history with the business, handles the replies, works the objections and books the appointment. This is the database reactivation offer most agencies sell, running natively
  • Rich elements on four channels. Quick replies, link buttons, lists, images, files and formatted text render natively on WhatsApp, Facebook Messenger, Instagram and Live Chat, and fall back to plain text everywhere else, so SMS and email still read correctly. Each element is switched on per agent under Response Behavior
  • Build with AI. Describe the agent in plain words and the whole thing is built in about a minute, triggers, channels, actions, hand-offs and working hours included, then refined by chat. Naming the business, the goal and the hand-off point produces a usable first pass
  • MCP servers give the agent your other apps. Connect Google Calendar, Drive, Sheets, Docs, Canva, Zapier or Apify from the Marketplace, or add your own server by URL, choose which tools the agent may call, and it looks up an answer mid-conversation instead of parking the question for a human
  • Claude and Gemini next to GPT. The Model menu now lists Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 and Claude Haiku 4.5 plus Gemini 3.8 Flash, Gemini 3.7 Flash and Gemini 3.5 Flash Lite, each with a relative cost indicator, so a high volume FAQ agent and a sales agent no longer have to run the same model
  • Better capture, correct times. The Contact Info action now writes Radio, Checkbox, Dropdown single, Dropdown multiple, Phone, Contact Source and Contact Type from what the customer says in their own words, and the agent works out a contact's time zone from their city before offering any slots, which ends the wrong-hour booking
  • Testing and deployment in one place. Test your agent previews replies the way each channel renders them and can test an outbound first message without texting a real contact, and the redesigned Deploy tab holds channels, outbound and triggers together
  • Workflow agents get a chat page. The AI Agent action in workflows can publish a hosted chat page with its own link, sign-in by one-time code against an existing contact, a display name, a welcome message and suggested prompts, which is an agent you can share without a website

Workflow Maps put every automation on one canvas

Workflow Maps is a new visual layer over the whole sub-account, showing workflows inside their folders alongside the tags, forms, calendars, pipelines and custom fields that connect them. It is generated from what already exists, so there is no diagram to maintain. Four companion releases make it an operational tool rather than a picture.

  • See how one thing starts another. A form starts a workflow, that workflow adds a tag, that tag starts another workflow. The map draws those chains, which is the account knowledge that normally lives in one person's head
  • Follow one contact's journey. Search a contact in the Contacts rail and see every workflow they entered with the runs and steps behind it, which answers why did this contact get that message without opening five workflows
  • Find errors across the account. An errors badge flags workflows with failed executions from recent runs, and Review workflows with errors groups them into published and unpublished, so a broken automation surfaces on its own instead of being noticed by a client
  • Linked workflows, and an assistant. Open a workflow together with everything it hands off to, including across folders, and ask the built-in assistant what a workflow does, where a problem is, or to build or edit one from a description
  • Workflow testing that is real. In Labs, test a single trigger with a real event from your own account, pin that event, then step through every action below it. The old test ran the whole workflow and hoped
  • AI Builder fills in your data. Name the form, pipeline, stage, snippet, tag or user in the prompt and AI Builder finds each record in the account and configures the step with it, so there is no manual pass after the build
  • An A/B Test action. Split contacts across 2 to 10 paths inside one workflow, choose what counts as a conversion, set when the test ends, then route new contacts to the winner or keep splitting. This is a split test on follow-up, not on a page
  • Slack and Google Sheets can start workflows. Five Slack triggers cover direct messages, channel joins, app mentions, channel messages and emoji reactions, and four Google Sheets triggers fire on a new spreadsheet, a new worksheet, a new row and an updated row

Email marketing was rebuilt around a prompt

Email Marketing got a new interface, a first visit that loads up to seven times faster, and an AI editor in Labs that turns a description into a send-ready email. Deliverability moved the same day, with a pre-send spam check and domain warm-up built into LC Email.

  • Email AI Editor. Describe the email and get a complete design in one pass, with hero image, headline, benefit sections, testimonial, calls to action and footer, already in your Brand Board colors, then refine it by chat, in code, with visual edits or in the design editor
  • Whole sequences from one prompt. Describe the goal, answer a few questions about audience, sender and start time, and Email AI lays out every step on the sequence canvas, sets the timing between them and writes a designed email for each step
  • A/B tests by asking. Once a draft exists, ask for a subject line or content test and Email AI configures up to four variations, the test size, the audience split, the duration and the winning criteria
  • Pre-send spam check. Run a check from the send screen that reviews content, sender setup and reputation signals and returns specific fixes, which is the step most senders only take after a campaign has already underperformed
  • Inbox Preview and mobile controls. See the email as Gmail, Outlook and Apple Mail actually render it rather than a simulated desktop preview, and style supported sections separately for mobile, including padding, margins, button alignment and full width buttons
  • Domain warm-up inside LC Email. Warm-up mailboxes are created on your dedicated domain and daily volume ramps toward a goal you set, with threaded conversations that build reputation at Gmail, Outlook and Yahoo. No third-party tool, no SMTP credentials and no extra subscription, and agency and sub-account domains are both supported
  • Two fixes that save a rebuild. A template switcher changes the template of an email already in the builder, with a confirmation before content is replaced, and contact import can verify email addresses as it runs, tracked under Bulk Actions. Email in HighLevel now covers the build, the test and the list hygiene in one place

The CRM learned how companies and records connect

Associations moved from a Labs experiment to the way the CRM works. Companies create and link themselves, contacts can belong to several of them under named roles, and those relationships are now filterable, triggerable and usable inside workflows.

  • Companies create themselves. Domain-based company creation left Labs for every sub-account. When a contact is created or their email changes, HighLevel matches the company by email domain, associates the contact, and creates and enriches the company when there is none. It is on by default for new agencies and off until an admin enables it for existing ones
  • One contact, many companies. A custom association label between contacts and companies lets consultants, shared finance contacts and owners sit against every account they touch without duplicate records, while Primary Company stays a fixed system label
  • Roles on the relationship. Labels such as Employee, Billing contact or Decision maker describe what a contact does at a company, with paired names like Employer and Employee, limits in each direction, more than one role for the same person, and email from a company targeted by role
  • Associations in Smart Lists. Every association label can have its own column, linked records show as chips with a plus count, and filters cover contains any, does not contain any, is empty and is not empty, so a segment can be built on relationships rather than on fields alone
  • Automation on the relationship. An Association Created trigger starts standard, company or custom object workflows the moment records are linked, filtered by object and label, and company and custom object workflows gained a Send SMS action that picks its recipients by association label
  • Calculated fields in private beta. A new field type fills itself from a formula over other contact data, which covers lead scoring, days since last engagement, countdowns to a renewal date and consistent formatting without a workflow
  • Record housekeeping. Documents on a contact can be filed into nested folders with 60 day recovery of deleted files, and the activity panel was rebuilt in Labs with cards for opportunities, appointments, notes and tags plus filters. The CRM is finally shaped for agencies with B2B clients

Forms, surveys and quizzes gained an Integrate tab and real order forms

The form builder picked up a dedicated Integrate tab, conditional logic on individual products, automatic tax calculation and the ability to create CRM records with no contact attached. Together they move forms from lead capture toward checkout and deal creation.

  • Everything about sharing in one tab. Embedding, live embed previews, ready-made QR codes, email options, workflow connections and post-save actions moved out of a modal into an Integrate tab inside the builder
  • A form can create an opportunity with no contact. A quote request or project inquiry creates the opportunity on its own, or creates two linked records: an opportunity and a company, an opportunity and a custom object record, or a company and a custom object record
  • Product-level conditional logic. Conditions can be set on individual products, and products and variants can be shown or hidden based on what the customer selected, which lets one order form serve several offers
  • Automatic taxes and separate billing addresses. Order forms calculate tax from the customer address and collect a billing address separately from shipping, which closes the main gap against a dedicated checkout. Selling through HighLevel now works from a form, not only a funnel step
  • Errors explained before publishing. Error detection names the field causing the problem and recommendations flag common mistakes while the form is still being built, which is one fewer support ticket per build
  • A native quiz element. Quizzes drop into funnels and websites as an element, with no copying an embed code back and forth between two builders

Brand Boards, a graphic library and QR codes that behave

The builders gained a native graphic library, and the Brand Board became something the rest of the product reads from rather than a reference sheet. QR codes, which HighLevel has expanded all month, picked up limits, new types and brand colors.

  • 11,000 icons, emojis and illustrations. A native graphic library sits inside the funnel, website, store and webinar builders with category browsing and search, recolorable icons, AI generated illustrations and your own uploads, attachable to an element or dropped on the canvas
  • Typography in Brand Boards. A full hierarchy covering Headings 1 to 6, Subtitle 1 and 2, Body 1 and 2, Caption and Overline, with smart font mapping, manual overrides that persist and a side by side preview
  • Brand boards without leaving the builder. Create one from the email builder's color picker in a modal, pulling colors, logo and fonts from a website URL or starting from a preset, and preview any template in your brand colors before you pick it
  • QR codes with scan limits and expiry. Set how many times a link can be scanned and how long it stays valid on a code that has already been printed, with limited codes badged in the list, alongside a rebuilt builder
  • QR codes for calendars, services and rentals. Booking and scheduling experiences get native QR types plus prebuilt codes inside calendars, services and rentals, and any code can take Brand Board or saved custom colors instead of plain black

Ad Manager added TikTok and two-way Meta sync

TikTok joined Meta, Google and LinkedIn in HighLevel Ad Manager, and existing Meta campaigns now sync both ways instead of sitting outside the platform. For an agency that reports from HighLevel, this is the gap that used to force a second login.

  • TikTok lead generation campaigns. Build, publish, manage and measure TikTok campaigns in the same builder as the other channels, collecting leads through existing TikTok Instant Forms or through Direct Messages with a greeting and suggested questions
  • Existing Meta campaigns appear. Connect the Meta ad account, choose which campaigns to sync by status and time period, and campaigns, ad sets and ads load into Ad Manager in the background
  • Edits flow in both directions. Change a campaign in Ad Manager or in Meta Ads Manager and the two stay in sync, so reporting in HighLevel covers all Meta spend rather than only what was built there. Accounts already connected turn two-way sync on in Ad Manager settings

Meeting Room, Voice Intelligence and a copilot in your pocket

HighLevel shipped its own video conferencing, automatic post-call notes and a mobile version of Ask AI Computer. Each one takes a tool out of the stack rather than adding a setting to an existing one.

  • Meeting Room. Built-in video with a permanent personal room link, instant meetings, automatic links on calendar appointments, guest access with no account, HD video and audio, screen sharing, live chat, co-hosts, microphone and camera controls, waiting rooms, passcodes, recordings, transcripts and AI summaries
  • Breakout rooms. Up to 100 rooms in a single meeting, with automatic or manual assignment, participant self-selection, host broadcasts, a session timer, extensions and automatic return to the main room
  • Voice Intelligence in Labs. Every recorded call produces what the contact wanted, what was agreed and what happens next, so the next person to call reads the summary instead of asking the customer to repeat themselves. Voice AI now covers the work after the call as well as the call
  • Ask AI Computer on mobile. The private cloud machine that browses in a real Chrome, fills in forms, runs code and works with files is now in the HighLevel mobile app, startable from any chat, watchable live full screen, with a tap to take over and hand back, and sign-ins routed through a private form and Secure Store rather than the chat
  • Badges for Communities. Members earn badges automatically from milestones, posts, comments, reactions and consistent participation, one badge shows beside their name, an almost there notification fires at 80 percent of a requirement, and the member list filters by badge
  • Conversations panels resize. On October 28 the dividers between the conversation list, the message thread and the contact panel became draggable, with each layout remembered per user

Affiliate Manager opened up, and multi-brand SaaS got its own sender domains

Sub-Account Affiliate Manager took four releases aimed at agencies running affiliate programs at scale, and Multi-Brand SaaS accounts can now give each brand its own notification email identity.

  • 32 public REST endpoints. Affiliates, campaigns, commissions, payouts, payout methods, referred customers and leads and affiliate media, with create and update access, standard v3 OAuth and two new scopes, so dashboards and CRM syncs can be built outside the interface
  • Denied PayPal payouts are visible. Payouts used to be marked paid without checking what PayPal returned. A Failed tab now shows denied and failed payouts with the reason attached, a Processing tab covers those still being confirmed, and every failed payout can be retried from the record
  • Audit logs for the module. Campaign, affiliate and customer changes are recorded with before and after values, filterable by action, affiliate, campaign, customer or payout, which is what makes a commission dispute answerable
  • Unlimited hiring depth. Affiliates at any level can recruit sub-affiliates regardless of how many commission tiers a campaign has, while commissions stay as configured and are counted back from whoever made the sale, so payout cost does not move
  • A notification domain per brand. Multi-Brand Management accounts can configure a separate notification email domain for up to two brands, each with its own DNS setup, so system emails come from the brand the customer actually knows. White-label SaaS loses another shared-sender giveaway

WhatsApp marketing has to move to MM Lite before October 2027

On October 19 HighLevel confirmed Meta's cutoff: from October 13, 2027 marketing messages can no longer be delivered through the Cloud API, and every WhatsApp marketing send has to run on Meta's Marketing Messages API, which HighLevel already ships as MM Lite. Twelve months is a long runway, so the deadline is not the reason to move. The reason is that a campaign still running on Cloud API goes out with no delivery optimization, no view of the funnel past read receipts and none of the creative formats Meta now keeps for MM Lite only.

  • Marketing sends only. Service and utility messages stay on the Cloud API exactly as they are today. Nothing in this notice touches transactional traffic
  • Cloud API is frozen for marketing. Meta says it stays available until the cutoff but receives no new marketing capability, so the gap between the two endpoints only widens from here
  • Delivery is scored on engagement. MM Lite weighs engagement signals when it decides what to deliver. Meta's own A/B test across more than 12 million messages measured up to 9 percent higher delivery for high-engagement messages, on the same send volume rather than a bigger one
  • Formats Cloud API will not get. Animated GIF template headers, Android app deep links and a pilot where Meta enhances template images for you. Media, carousel, product catalog, flows and interactive lists and replies all keep working
  • Measurement goes past read. MM Lite reports the funnel from message through to purchase, plus spend per template, cost per click and cost per delivery, so a WhatsApp campaign can finally be priced per outcome instead of per send
  • Nothing gets rebuilt. Existing phone numbers and approved marketing templates carry over with no re-creation and no re-approval, campaigns keep running through activation, and sub-accounts on the WhatsApp Business App can coexist
  • Activation takes minutes per sub-account. In the sub-account open Settings, then WhatsApp, where an activation banner appears. Enable MM Lite sends you to the WhatsApp account interface, where the Requests tab has a Respond button, then Continue, then Finish on the Review LeadConnector's access request screen

Start with the sub-accounts sending the most marketing volume, because those collect the delivery and measurement gain first. Our WhatsApp features overview covers what the channel does inside HighLevel either way.

WhatsApp gained carousels, in-chat forms, call recordings and a template library

On October 26 five WhatsApp releases landed together, and they change what the channel can carry rather than how it looks. A marketing message can now hold ten products, a form can be answered without leaving the thread, and every call on a WhatsApp number can be recorded and written up. It is the same week HighLevel told everyone to move marketing sends to MM Lite, so the channel is being built out and repointed at once.

  • Carousel templates hold up to ten cards. Write the body once, then attach 2 to 10 cards, each with its own image or video header, optional text and up to two buttons, and the customer swipes through them in the chat. WhatsApp bills a carousel as one marketing message and counts it once against your frequency limits, so ten offers cost what one used to. A tap on card three tells you which product landed, quick reply buttons answer in the thread and fire a webhook, and each card's website button can carry its own variable. Card count is locked at approval, and every card has to match the others on body text and button count
  • Forms open inside the conversation. Send a form you already built as a WhatsApp Flow and the lead answers it in the chat window, with no link, no browser and no landing page to abandon. Responses land in Forms, then Submissions, mapped to the right contact. Field types WhatsApp Flows cannot render are stripped at build time rather than discovered from a broken send, and a published flow keeps collecting what it was published with even if the underlying form is edited later
  • Calls are recorded and transcribed. Turn it on once for a number and every WhatsApp call on it is captured. The audio sits in Conversations next to that customer's messages, with a transcript beside it separating your team from the customer, and the language of the call is picked up automatically. Callers hear a short announcement first, in a language and wording you set, so consent happens before recording starts
  • A library of 52 pre-built templates. 32 marketing and 20 utility templates written for home services, web design, legal, beauty and wellness, insurance and property inspection, in English (US), English (UK), Spanish, Portuguese (BR), French and German. Browse, search, filter by category, tag or language, sort by most popular, favourite the ones you reach for, then edit the name, header, body and variables and submit for Meta approval as normal
  • Booking flows in 13 languages. Appointment booking flows now carry a language setting, and the timezone picker, date and time selectors, field labels, helper text and confirm button all follow it. English, German, Spanish, Italian, Dutch, Norwegian, Danish, Finnish and Swedish, plus French for France and Canada and Portuguese for Brazil and Portugal. Form flows render in whatever language you wrote them in, the live preview renders in the chosen language before you publish, and language shows as a column in the flows list

The carousel is the one worth planning around, because it is the first WhatsApp format that puts a range in front of a customer for the price of a single send. Paired with the per-contact media headers from October 15, one approved template can cover a catalog. Our ecommerce overview covers what the store side of that looks like.

The October 13 voice price rise was cancelled

On October 12, one day before it was due, HighLevel cancelled the voice pricing update. The October 1 notice had repriced inbound minutes in 19 countries and outbound minutes to 21 destinations, all effective October 13. The new notice says the underlying providers confirmed the cancellation, so the affected voice rates stay exactly where they are. The messaging and phone number rental changes announced on September 21 are a separate notice and still take effect on October 13.

  • Voice rates are unchanged. Nothing to do and nothing to reset. Local and mobile incoming minutes keep their current list prices, as do all 21 outbound destinations. If you already raised a resale price for October 13, put it back
  • Messaging and numbers still move on October 13. Outbound SMS rises for twelve destinations and monthly number rentals rise across eleven countries. Hong Kong local numbers are the outlier, going from $6.00 to $14.70 per month, which is a 145 percent increase on one line item
  • What the cancelled notice would have done. Local incoming minutes were going to $0.0115 per minute across fifteen countries, mobile incoming to $0.0125 across seven, the United Arab Emirates toll free incoming rate from $0.3675 to $0.4220, and Morocco mobile outbound from $1.0748 to $1.2354 per minute. None of that applies now
  • Read the first invoice after October 13 anyway. Two notices sharing one effective date, with one of them withdrawn the day before, is how a wrong line item goes unnoticed. Check voice minutes against September and check SMS and rentals against the September 21 table

Our HighLevel pricing guide covers the plan tiers these usage rates sit on top of, and Voice AI runs on the same per-minute telecom rails, so the reversal reaches it too.

Reviews AI got brand controls, and reputation reached the dashboard

On October 26 Reviews AI agents gained brand voice settings, quality guardrails and a live response preview, agencies gained bulk control over Reviews AI across sub-accounts, and Location Dashboards gained 16 reputation widgets. Three releases on one day, all answering the same complaint. Automated review replies were hard to trust and harder to prove.

  • Brand voice is configurable. Each agent can answer as we or I, carry the business name, take tone traits, respect a response length and character limit and sign off the way the brand signs off
  • Guardrails cut the risky sentence. Agents can remember recent responses, avoid repeating the same opening line and refuse refund promises, guarantees, admissions of fault, liability statements and medical or financial advice
  • Sensitive reviews can wait for a human. Higher-risk reviews are held for review before anything is published, so the reviews that actually carry risk are still answered by a person
  • Preview before saving. The configuration screen shows how the agent answers a positive, a neutral and a negative review, and the samples regenerate as tone and prompts are adjusted
  • Agencies set the mode in bulk. Agency, then AI Suite, then AI Setup sets Reviews AI to Off, Suggestive or Auto responses for one sub-account or many at once, and a sub-account changing its own mode is reflected back in the agency view
  • Reviews AI can be switched off entirely. Agency, then Sub-Accounts, then Features and Limits disables it for a sub-account, which hides the settings, AI Reply and Regenerate and blocks every AI review reply including drip campaign replies
  • Reputation has 16 dashboard widgets. Requests sent, requests by channel and by status, total reviews, reviews by rating, source and sentiment, average rating, response rate, average response time, the trends for each and an AI versus manual response split, filterable by date range, rating, source and response status

The widgets are the part agencies will reach for first, because reputation work has never had a client-facing number attached to it. Our reputation management overview covers the rest of the module, and white-label setup covers what a sub-account sees.

Social Planner gained listening, client invites and comments in the inbox

Also on October 26, Social Planner picked up Social Intelligence in limited beta, a password-protected invite that lets clients connect their own social accounts, and comment sync into the Conversations team inbox. The existing Social Listening experience was not removed. It now sits under a Trends tab inside the new workspace.

  • Social Intelligence is a limited beta. Marketing, then Social Planner, then Social Intelligence. Overview reports brand mentions, industry trends, share of conversation against up to three competitors, post ideas and an AI summary. Explore searches brands, keywords, hashtags and topics for mention volume, sentiment and where conversations happen
  • Saved topics are where the depth is. Save a search as a topic to get 7, 15 or 30 days of history on engagement, conversations, posts and audience, and pause or resume a topic at any time. Overview itself only looks back 7 days
  • Beta limits are real. Each location gets 100 searches for the entire beta period, up to 50 topics can be monitored at once with paused topics excluded from that count, and each topic takes up to three alerts. A dedicated Competitors tab is listed as coming
  • Alerts watch the metric, not the feed. Mentions, sentiment, engagement, impressions and share of voice each take an alert with its own sensitivity and recipients, delivered daily or weekly by email
  • Clients connect their own accounts. Social Connect Invite, enabled under Settings, then Labs, generates a link with an auto-generated or custom password. The client connects on an external page and the accounts appear in the right sub-account, with no sub-account access handed over and no credentials shared
  • Invites are managed in one place. Settings, then Manage Links, then Social Connect Link extends, regenerates or revokes an invite. Links expire 30 days after creation, revoking is immediate, and anyone holding both the link and the password can connect accounts to your sub-account, so treat the pair as a credential
  • Comments land in the team inbox. Comments on posts published through Social Planner appear in Conversations with the commenter name or handle, a preview, the platform, a timestamp and the original post with a View Post link, and a Social Comments filter to find them
  • A commenter becomes a contact. Every new commenter is created as a contact and a repeat commenter is matched to the existing record, so the inbox does not fill up with duplicates
  • Public reply or private DM. Reply underneath the comment or open a standard DM thread, which is the move from a public interaction to a one-to-one sales conversation without leaving HighLevel
  • Intent can be prioritized. Buying-intent keywords such as price, interested or how much, and comments that contain questions, can be pushed to the top of the queue

Comment sync is controlled by the agency under Agency Settings, then Social Comments, and per platform in the sub-account under Social Planner settings, then Comments. Facebook, Instagram and TikTok are live at launch, with LinkedIn, YouTube, Pinterest, Threads, Bluesky and Community listed as coming. Only posts published through Social Planner sync, so a post made natively on the platform is still invisible here, and turning sync off stops new comments without deleting the contacts, conversations or workflows already created.

Conversations finally has keyword search, plus six new filters

October 26 also brought keyword search to Conversations, across the whole inbox and inside a single thread, along with six new filters. HighLevel calls search one of the most requested items on its Ideas Board, which is a fair description of a messaging inbox that could not search messages.

  • Search across every conversation. The search icon at the top of the conversation list matches any word or phrase as you type, with a total count and a preview of the matching message per contact, keyword highlighted. Clicking a result opens that conversation at the match, and the same box still accepts a contact name, email or phone number
  • Search inside one thread. The search icon in the conversation header highlights every match, counts your position, for example 6 of 6, and steps between matches with up and down arrows. Arriving from an inbox-wide search carries the keyword with you
  • Results narrow by filter. Date range, channel across email, SMS, Facebook, WhatsApp and Instagram, direction, owner, has attachments and followers
  • Opportunity filters bring the pipeline in. Filter the inbox by opportunity pipeline, stage and status, so open deals waiting on a reply can be worked without leaving Conversations. A contact with opportunities in several pipelines appears if any one of them matches
  • Channel Account filters by number or page. Pick the specific phone number, Facebook page or Instagram account the most recent message used, which is the filter agencies running many numbers have been asking for
  • Last Call Status and Contact Type. Filter by the outcome of the most recent call, completed, no answer or voicemail, and by Leads, Customers or any custom contact type in the sub-account. All six new filters combine with the existing ones using AND and OR

Between search and the opportunity filters the inbox stops being a list and starts being queryable, which removes most of the jumping between Conversations, Contacts and Opportunities. Our CRM overview covers how the pipeline and contact data these filters read is set up.

Social Planner posts can now trigger workflows

On October 25 a Social Planner Posts trigger arrived in Workflows, so publishing activity can start automation on its own. Until now a post that failed to publish sat there until somebody opened the calendar and noticed. The trigger fires on post status, which puts approvals and publishing failures on the same footing as every other event in the account.

  • Six statuses to fire on. Approval Requested, Approved, Approval Rejected, Scheduled, Published and Failed, so one workflow can chase an approval, confirm one, or raise an alert the moment a post does not go out
  • Failed posts stop being a silent problem. A Failed trigger plus an internal notification means somebody knows in seconds rather than at the end of the week, which is the difference between a reschedule and a missed campaign
  • Filter by category, platform and post type. Promotions, announcements and educational categories can each run different rules, platforms cover Facebook, Instagram, Bluesky, Google Business Profile, LinkedIn, Pinterest, Threads, TikTok and YouTube, and Carousels, Posts, Reels, Shorts and Stories can be handled separately
  • Target named accounts and tags. Connected Social Planner accounts and existing post tags both narrow the trigger, which is what makes it usable in an agency running several brands with a different approval rule per client
  • Where it lives. Automation, then Workflows, add the Social Planner Posts trigger, set the filters and publish. Actions behave as they do anywhere else, so internal notifications, CRM updates and follow-up steps are all available
  • One connection caveat. Only accounts connected through Social Planner are supported. Accounts connected through Integrations do not fire the trigger yet, and HighLevel says support is in progress

Approval chasing is the part of social media management that never gets automated, because it lives in chat rather than in the CRM. This moves it into the same automation layer as everything else, and the Social Planner mobile guide covers the posting side of the same tool.

HubSpot imports can be filtered, and in-app bookings can carry a form

On October 22 the HubSpot importer gained date filters on its preview step, and service and rental bookings created inside the app gained a booking form. Neither is a headline feature. Both close the same kind of gap, where records arrive in bulk and the cleanup lands on whoever opens the account next.

  • Pick which HubSpot records cross over. Contacts, Deals, Companies and Tasks filter separately on the Preview and import step, by HubSpot Created Date or HubSpot Last Modified Date, so a client moving off HubSpot brings the active records and leaves the rest behind
  • Relative or exact ranges, combined. In the last 6 months and This quarter sit alongside On, Before, After and Between, the exact dates a filter covers are shown before it is applied, conditions nest with AND or OR, and Apply filters to all objects reuses one set across every selected object
  • The count updates before you commit. Records found and records that will import both refresh as the filters change, so the size of the import is known before Confirm and start import. Filters are optional and an empty filter imports everything, and In the last excludes today
  • Bookings made by staff collect the same intake as bookings made online. Service and rental appointments created from the in-app modal have a Booking form section with an Attach form button, the configured form preselected and any other available form in the dropdown
  • Contact details prefill and the storefront elements drop out. First name, last name, email and phone come from the selected contact, and CAPTCHA, product selling and payment elements do not appear when a team member fills the form on the customer's behalf
  • One limit to plan around. A form can be attached only while the booking is being created, not added or edited afterwards, so a booking saved without one stays without one. The submitted form is readable later from the appointment details

Manual bookings and migrated records are the two places where data quality quietly degrades. Filtering one and standardizing the other removes work nobody schedules.

Call tracking learned which ad platform made the phone ring

On October 16 number pool calls started naming the ad platform behind each call, and a pool can now ring up to six named users on whatever device each of them answers. Call tracking used to file most paid Facebook calls as Facebook or Direct, which made the paid and organic split guesswork. Those calls now arrive labeled Facebook Paid, and TikTok, LinkedIn, Microsoft Bing and in-app Google clicks each get their own platform label.

  • Paid calls are separated from organic ones. Facebook ad calls read Facebook Paid instead of Facebook or Direct, so a call report can answer whether the ad spend or the page produced the call
  • Tagged links are read too. A link carrying utm_medium=cpc resolves to the platform plus Paid, so a Bing ad lands as Bing Paid. Where a link carries parameters from several platforms the first match wins, checked in the order Google, Facebook, TikTok, LinkedIn
  • AI referrals get their own source. Visits arriving from ChatGPT and comparable assistants are reported as themselves rather than folded into Direct, which is the first time a HighLevel call report counts an AI referral as a channel of its own
  • Nothing to install, and nothing retroactive. The number pool tracking script updates itself. The new labels apply to calls from October 16 forward and past calls keep the source they were already given, so a channel comparison that spans that date is not like for like
  • Direct does not disappear. A visitor whose browser withholds the referrer, on a link with no ad tracking parameters, still cannot be labeled. Treat a residual Direct bucket as expected rather than broken
  • A pool can ring a team instead of a line. Ring Incoming Call to selected Users in the number pool settings takes up to six team members, rings all of their devices at once and remembers the selection next time the pool is opened. Each person answers on web, mobile, a desk phone or their own number, following the Default channel for ring all set under Settings, then Users or My Staff, then Edit, then Call and voicemail

Both changes land in the same place, which is the point. A tracking number that reports the platform and rings the people who should answer turns call attribution into a report you read rather than one you reconstruct.

WhatsApp templates were rebuilt, with media that changes per contact

On October 15 the WhatsApp template builder was replaced and media headers stopped being static. Category, type, naming, body, media, buttons and variables now sit in one redesigned builder with a live phone preview, and a media header can point at a custom field instead of a single hard-coded URL.

  • Dynamic media headers. A media header field accepts custom fields and dynamic values. At send time the value resolves for that contact and Meta receives a concrete URL for that one message, so every recipient can get their own invoice, ticket or image from one approved template
  • One template instead of a library. A different document used to mean a different template, and every template goes to Meta for approval. A single approved template now serves every contact's asset, which takes the approval queue out of the loop
  • It works wherever you already send. Workflows and automations, conversations, bulk actions, calendar appointments, reputation review requests and missed call text back all support dynamic media headers
  • Live phone preview while you build. The preview updates as you edit the header, body, media and buttons. Catching a layout problem in the builder costs seconds. Catching it after Meta approval costs another approval cycle
  • One builder everywhere. AI template generation runs through the same builder end to end, and the WhatsApp template builder opened from Reputation uses it too, so there is one interface to learn rather than three

The same day a limited time offer template type arrived, with a countdown WhatsApp draws itself. Create it under the Marketing category with an offer heading, an image header and a body with variables. The clock ticks down inside the customer's chat until the expiry you set and turns red in the final hour, so the deadline is rendered by WhatsApp rather than claimed in your copy or baked into an image you redo every campaign.

  • The expiry is set at send, not at build. The template defines the structure and each campaign picks its own deadline, so one approved template runs promotion after promotion with no wait on Meta approval between them
  • Copy code and URL buttons. The discount code sits under the clock and copies with one tap, with a URL button beside it going straight to the offer. Meta requires both buttons on this template type, so neither can be removed
  • Send it from everywhere you already send. Conversations, bulk actions, workflows and calendar sends all support it, and the AI template writer can draft one including the offer heading, expiry, offer code and image
  • One thing to watch. Meta still accepts a send carrying an expiry that has already passed, so check the date on every campaign rather than trusting the template to catch it

Both releases point the same way. WhatsApp in HighLevel is moving from one template per message to one template per pattern, with the per-contact detail filled in at send time.

Workflows can post internal comments in conversations

On October 14 a new Add Internal Comment action landed in workflows. Internal comments were manual until now, so the context the next person needed got typed by hand or kept somewhere else entirely. The action writes the comment into the conversation history itself, and nothing reaches the customer.

  • Where it lives. Automation, then Workflows, then add the Internal Comment action to a new or existing workflow, write the comment, save and publish
  • It can @mention users. Tag the assigned user or a named team member inside the comment, so the instruction reaches the right person where the conversation already is instead of in a separate tool
  • What it is for. A high priority lead can arrive carrying its own handover note, a stage change can attach the next step automatically and routing context stays attached to the thread
  • Internal only. The comment is visible to staff in the conversation and is never delivered to the contact, so account history, pricing notes or a warning are safe to put in it

WordPress hosting got antivirus, one menu and comments on blog posts

Also on October 26, every WordPress site gained a security scanner, a rebuilt navigation and a sites table that says which site needs attention. Blogs gained reader comments the same day. None of this is new surface area to sell, but it closes the gap between hosting a client site and being able to say something about its condition.

  • Antivirus on every site. Shield scans for harmful files and common security gaps, lists each problem and why it matters, and fixes them with Apply fix or Apply all fixes. Ten checks return a clear Passed or Action required, covering core and plugin files compared against their official versions, unused plugins, switched-off protections, HTTPS, the WordPress version, supported server software, hidden error messages and search engine visibility. An item can be snoozed until the next scan, for 7 days or for 30, and the last 30 days of fixes stay in Resolved
  • Ten one-click hardening toggles. Block application passwords and user enumeration, hide the WordPress version, mask login errors, disable the file editor and XML-RPC, add security headers, rate-limit login attempts and harden .htaccess, with Turn all on for the lot. Leave XML-RPC enabled if the client uses the WordPress phone app
  • Threats across every site. A Threats column on the WordPress sites list shows a count per site, hovering it shows what was found, and every site can be scanned in one click
  • One menu per site. Each site opens to a single page with tools grouped by job: Info for live access, staging and domains, Analytics, WordPress Management for plugins, themes, core, pages and users, Backups, Antivirus and Advanced Settings. Site name, address, Site ID with a copy button and region sit at the top, and the arrows beside the site name search and jump to any other site without going back. Troubleshooter and Activity Logs moved under Antivirus
  • A sites table that reports. Status, Threats, Updates, Uptime, Backups, Performance and Storage for every site, one clear action per row such as Manage site or Finish setup, search, paging and a list or cards view
  • Comments on blog posts. Add the Comments element to a blog post page once and every post on that blog site gets a comment section, with replies, emoji, one image per comment and likes. Readers sign in with an email and a one-time code, no password. A Comments tab shows total comments, comments this month and comments awaiting approval, each with a trend line, plus filters, search and reply, approve or delete on any comment. Team replies carry a Moderator badge, and settings cover required approval, an email per comment or a daily digest, who receives them, and applying the same settings across every blog site

Two limits on comments matter before you switch them on for a client. Comments are per blog site rather than per post, so removing the element removes them from every post, and comments do not reach Conversations yet, so commenters are not created as contacts. On the security side, every fix changes the live site, which is why Antivirus shows the date of the last backup and offers Back up first.

Sites gained a schema dashboard and real checkout numbers

On October 12 schema markup moved out of the page builder. Every website and funnel now has an SEO and AEO tab holding a schema dashboard for the whole site. Keeping schema current used to mean opening each page in the builder one at a time, which is why most sites drifted out of date.

  • A coverage score for the site. The Schema coverage card rates the site Needs work, Getting there, Good or Excellent, with counts for Up to date, Might be outdated and No schema
  • Every page on one screen. Each page lists its schema types, its status of Up to date, Might be outdated, No schema, Unpublished or Unsaved, and when its schema was last updated. Search by page name or path and filter by status
  • Bulk generation with review. Select pages, click Generate schema, watch progress live, then read the output before it goes anywhere. Generated or edited schema stays Unsaved until you click Save and publish, and Discard throws it away
  • Manual control is still there. Add schema on a page that has none, or click any schema tag to open the editor. Editing inside the page builder works exactly as before and both views stay in sync
  • Stale schema gets flagged. A page edited and republished two or more times since its schema was created is marked Might be outdated, which is the check nobody was running by hand
  • One trap. Running Generate schema on a page that already has schema replaces it entirely, so review before you save, especially where the markup was written deliberately

Nothing to switch on and no data loss. Existing schema appears in the tab automatically and pages without any are flagged so they can be fixed in bulk. Accurate markup is what makes a page eligible for Google rich results and what helps AI answer engines read and cite a business, so this is the first schema tool in the platform aimed at a site rather than a page. Run the Might be outdated filter first, then generate for the No schema list.

The same day, Meta Pixel tracking stopped counting checkout twice. InitiateCheckout now fires when a buyer reaches the checkout page rather than when they submit payment, and AddPaymentInfo carries full order and customer detail. It applies to funnels, websites and stores that have the Meta Pixel set up with the InitiateCheckout event enabled.

  • When it fires now. On page load for store checkout pages and for any page where InitiateCheckout is set at page level, and on the buyer's first click, tap or keystroke in an order form everywhere else. One-step, two-step, popup and the new order form are all covered
  • No double counting. It fires at most once per page visit even where a page holds several order forms or more than one trigger, and browser autofill, password managers and fields focused automatically on load do not set it off
  • Abandoned carts and express wallets. A buyer returning through an abandoned cart link triggers the event once the saved cart has loaded, so the cart value and products are included. A buyer who goes straight to Apple Pay without touching the form triggers it when payment starts
  • Richer AddPaymentInfo. Order value, currency, product IDs and hashed email, phone, name and external ID where available, which is what Meta uses to raise Event Match Quality and improve attribution
  • Expect the counts to rise. InitiateCheckout now includes buyers who reach checkout and leave, so the gap between it and AddPaymentInfo is real drop-off for the first time. Any report or automated rule built on cost per InitiateCheckout needs a new baseline
  • Timing note. Page-level settings can take up to 24 hours to apply on pages that are already published. Editing and saving the page applies them at once

Consent handling and the Conversions API are unchanged. Nothing is sent when a visitor has not accepted advertising cookies, and browser and server events are still sent together and deduplicated in Events Manager. Free orders still skip AddPaymentInfo, with one change: an order fully covered by a gift card now sends it with a value of 0. If InitiateCheckout is not enabled on your pixel, nothing about this release reaches you.

AI Studio became a real app platform

In two days AI Studio stopped being a page builder. Three releases landed. Apps built in AI Studio can take payments through HighLevel's own checkout, and every published AI Studio site now reports its own traffic and opt-ins with no tracking code to install. Since October 9 each project can also run a private backend of its own.

  • Payments live in a settings tab. More, then Payments. Switch payments on, pick test or live mode, and choose which catalog products the app is allowed to sell. Each product is priced as one fixed price, a chosen set of variations, or its full catalog range, and only what you allow can be shown or charged
  • The agent builds the checkout for you. Ask the agent to add payments, approve the connection on the card it shows, and it builds a storefront and checkout into the app. The payment code is managed by the platform, so edits to it are restored on the next turn rather than quietly breaking the cart
  • A hosted checkout, not your code. Card details are entered and processed in the platform's hosted checkout. PayPal express is supported, coupon codes work, a fully discounted order goes through without asking for a card, subscriptions work including carts holding more than one subscription product, and the checkout picks up your site's color scheme instead of a fixed default
  • One-click post-purchase upsells. An offer page after a successful purchase charges a saved card with one click, on Stripe with 3D Secure, Square, Authorize.net, NMI, Adyen and Mercado Pago. Other payment methods fall back to the regular form. An upsell charges only after the original order is paid, and only once per order even with several tabs open
  • Analytics with nothing to install. More, then Analytics. Page views, unique visitors, opt-ins and opt-in rate with trends and the change on the previous period, plus top pages, time on page, traffic sources, devices, browsers and geography. Form Submissions lists every lead with a link to the contact, filterable by form and date, and a workflow can be built from there
  • Visit history lands on the contact. Contacts created from an AI Studio site keep the pages they viewed, where they came from and the form they submitted, under Contacts, Contact details, Activity. Sites, then Analytics, then AI Studio puts every project next to your funnels and websites

Two caveats decide whether this reaches you this week. Payments works only on apps built on TanStack Start, so an existing React project needs Project Settings, then Migrate to SSR, which makes an upgraded copy and leaves the original live while you test. Forms connected before October 8 keep counting opt-ins, but linking a submission to the pages that visitor saw first needs a reconnect and a republish. AI Studio left Labs on September 22, so this is a general release rather than a beta, and the ecommerce guide covers the catalog, orders and payment providers these checkouts run on.

On October 9 AI Studio added the part that was missing: a backend. More, then Cloud, then switch it on from Overview. About a minute later the project has a private database, user accounts and file storage with nothing to connect. Setup runs in the background so you can keep building, a failed setup leaves the app untouched and Try again resumes it, and AI Studio asks before it builds anything against Cloud.

  • A database you can actually open. Every table with its approximate row count and whether row-level security is on. Add rows, double-click a cell to edit it, select rows to delete them. Ask chat for a new table or policy and it writes the schema. Optimize sends a health check into chat and costs credits like any message. Daily Backups snapshot schema and rows, not storage files or app code, and Restore swaps the live database for a snapshot while database access pauses for a few minutes
  • User accounts without an auth provider. A signup chart, search by email, Create new user for someone already confirmed, Send invitation for an email link to set a password. Signup, invite and reset emails go out through your account's email and the links open your app. Email confirmation is required by default, Allow new sign-ups closes registration without signing existing users out, passwords found in public breaches can be rejected, and Site URL and Redirect URLs decide whether a login works at all. Version one is email and password only
  • File storage in buckets. Public buckets open to anyone with the link, private ones only to signed-in users. Bucket names are lowercase letters, numbers and hyphens, 3 to 63 characters, and permanent. Upload, folders, download, rename, delete and Copy URL, with an upload size limit and RLS policies controlling read and write. Renaming a file changes its URL
  • Hourly pricing you set yourself. Database size starts at Tiny, which is free, and moves up to Mini, Small, Medium, Large or X-Large, or back down, once an hour, with each size showing its hourly price. Database storage grows but never shrinks, up to four times a day, and bills for the space reserved even while paused. File storage bills for what is actually stored. Pause stops access and keeps the data, and a wallet that cannot cover Cloud pauses it rather than deleting anything
  • A security advisor on the Overview. When the advisor finds a problem a card appears, View issues lists them and Try to fix hands them back to AI Studio. Critical issues are worth fixing before you publish, warnings are worth a look

Three things to know before switching it on. Cloud needs an app built on TanStack Start, the same requirement as payments, so a React project goes through Project Settings, then Migrate to SSR first, and migrating does not turn Cloud on by itself. The published site stays as it is until you publish again. Sharing or remixing a project does not copy its Cloud data, and the copy can still send requests to the original project's Cloud under that project's access rules. Remove Cloud permanently deletes the project's database, users and files, so export tables as CSV and download the files first.

On October 14 AI Studio apps gained a chat widget of their own. Ask AI Studio to add a live chat and it creates a real HighLevel chat widget, names it after the project and wires it into the app. The chat is built into the app's own code, so it takes your site's colors, fonts and layout instead of dropping a generic bubble on top of them. Two types are supported, Live Chat and Email Chat for visitors who leave a message and get a reply by email, and one project can run more than one widget, for example live chat on every page plus a separate after-hours chat.

  • The real widget underneath. Visitor sessions use secure tokens that never reach the app's code, a captcha appears when the chat suspects a bot, rate limits and your spam settings apply to every message, visitors can attach files within the widget's limits, returning visitors pick up where they left off, and messages show as sending, sent or failed with a retry. Business hours, out-of-hours messages, inactivity timeouts, chat ratings and chat-ended messages follow the widget settings you already have
  • Every message lands in Conversations. Exactly as the standard widget does, so the threads and contacts are the ones your team already works in rather than a second inbox to watch
  • Chat analytics of its own. More, then Analytics, then Chat Widgets. Chat loads, opens, open rate and visitor satisfaction from chat ratings, plus a table of every lead with contact, email, phone, chat type, the page the chat started on and the widget it came from. Filter by widget and date range, choose the columns, select leads to download
  • Widget settings without leaving the builder. More, then Chat Widgets. Rename a widget, edit the welcome and acknowledgement lines, set the contact form fields, add compliance notices and opt-in consent for privacy, marketing, SMS and HIPAA, keep suspicious leads for review with optional DND on SMS or email, and set live chat business hours with an offline message outside them

This is the first AI Studio release this month that needs no migration: chat widgets run on React projects as well as TanStack Start, unlike payments and Cloud. Voice chat, WhatsApp, Facebook Messenger and Instagram widgets are not supported in AI Studio yet, so those conversations still start from the main app.

On October 15 AI Studio gained two-way GitHub sync. Connect a GitHub account to a project under More, then Git, as a personal account or an organization, shared with the sub-account or kept to yourself. A private repository named after the project is created with the first push already done, and reconnecting a project reuses its old repo rather than force-pushing over it.

  • Both directions, automatically. Every version you save is pushed, whether it came from a prompt, a code edit, a visual edit or a restore, and new commits on the connected branch are pulled back. Either direction can be switched off, and Push now and Pull now are always available
  • Build in whatever editor you like. Work in an outside editor or coding agent and the changes land back in the project. A fresh preview builds in the background and waits behind Update preview, so the page never reloads under you, and a failed build offers Retry build or Try to fix
  • Diverged branches ask rather than guess. When GitHub and AI Studio both hold new commits you choose Force push, where AI Studio wins, or New branch, which keeps both. Sync History shows total syncs, success rate, the push and pull split and every commit that moved, each linked on GitHub
  • Secrets block the push. Pushes are scanned for AWS, GitHub, Stripe and OpenAI keys and for private keys, and a match stops the push instead of publishing it. Images, fonts and other binaries survive every sync
  • A new permission, on by default. Manage Git and download code controls the Git tab and code downloads for every role, and an agency admin can switch it off per user under Settings, Team, Permissions, AI Studio. Without it a user still builds, edits and previews as before

This one needs no migration either. Git sync runs on React projects built with Vite as well as TanStack Start, only the connected branch syncs, and commits made by the platform are authored under its own identity rather than yours. A project that never connects GitHub sees no change at all.

The same day the AI Studio checkout became configurable. Payments shipped on October 8 with a fixed form. Three changes landed on October 15, and a project already using Payments picks all of them up on its next prompt with nothing to reconnect.

  • Choose the fields. Phone and each billing address field, street, city, state, ZIP and country, can be shown, hidden or required. Name and email stay required. Ask the agent for it in plain words, for example make phone required and hide the billing address
  • Two-step checkout. Step one saves the buyer as a contact and step two takes the payment, the same shape as the two-step order form in funnels, so a buyer who leaves before paying is still a lead you can follow up. Check the trigger: those leads fire Contact Created, not Order Form Submission, so a workflow listening for the wrong one silently never runs
  • Order details after payment. The confirmation screen shows the order ID, the items, taxes and total paid, plus the next charge on a subscription, or buyers can be sent to your own thank-you page showing their real order

The Lead Source Report stopped being a counter

On October 6 the Lead Source Report widget learned what each channel is worth. Until now it counted leads by source and put a single value on the total. It now prices every status and lets you click through to the records, which is the difference between knowing a channel is busy and knowing it pays.

  • Value for every status. Open, Won, Lost and Abandoned each carry their own monetary value on the source row. Compare won value against lost value per channel and the spend question answers itself
  • Drill down from any number. Click any count greater than zero under Total leads, Open, Won, Lost or Abandoned and the exact opportunities behind it open. No separate search, no rebuilding the filter by hand
  • Attribution gaps are visible. Opportunities with no source appear as an N/A row instead of being left out of the report, and the Source filter accepts Is empty and Is not empty. A report that quietly dropped unsourced records was flattering every channel in it
  • Lead value permissions still hold. The figures show only to users who can view lead value, so the drill-down does not hand revenue numbers to staff you scoped out of them

Row heights and column widths were tidied in the same release, so the table survives a laptop screen. HighLevel lists support for a custom field as the revenue source as what comes next. Run the Is empty filter first. If a meaningful share of your opportunities land in the N/A row, every channel comparison you have made this year was built on a partial table, and the CRM guide covers where sources get set in the first place.

A2P registration stopped being a black box

On October 6 two smaller releases landed next to the Lead Source Report, and both remove a guessing game.

  • Live A2P registration status. Every number attached to an A2P campaign now shows Pending, Registered or Failed, and a number fully registered with carriers carries an A2P Verified tag on the phone numbers page. Existing numbers were updated automatically
  • Error 30034 becomes avoidable. A number previously showed only that it had been submitted. Carrier registration takes anywhere from minutes to several days, and anything sent inside that gap failed with error 30034 and usually a support ticket. You can now see the moment a number is ready to text, and spot a rejected registration without asking anyone
  • Rejection guidance that matches how you registered. On October 8 the View required fixes modal was rewritten. Common rejection reasons now carry separate instructions for Chat Widget quick setup and for manual setup, because the fix sits in a different place for each. Where carriers require exact wording, the steps give it. Where the cause is something the platform generates, such as the chat widget's consent text, it says so and points you to support instead of spending a resubmission, and where a resubmission cannot help, such as a business type mismatch, it says what has to change first
  • One confirmation per recurring booking. Default recurring appointments sent a confirmation for every occurrence. They now send one that covers the whole series, while reminders and follow-ups still fire per occurrence
  • Custom recurrences get notifications at all. Custom recurring appointments created in-app previously sent nothing. They now send confirmations, reminders, follow-ups, reschedules and cancellations, following the This appointment, All appointments or This and following option you pick when you make the change
  • The pattern shows in the message. Custom recurrence notifications can print the recurrence next to the date, for example October 14, 2026 12:00 PM (reoccurs every Thursday), so a recipient can tell whether a change hit one session or the series. Your template needs the appointment date and time custom value for that to appear, and this covers calendar notifications only, not workflows

Registration usually completes within 24 hours but the timing belongs to the carrier, so a Pending status inside that window is expected rather than broken. Check the pill before the first send from any number you bought this month. The appointment changes need no setup and apply to newly booked appointments and to any appointment you edit.

Ask AI gained a voice and invoices gained a designer

On September 25 Ask AI in the mobile app learned to hold a spoken conversation. Tap the mic and speak. Your words transcribe live, the reply streams back as audio with the text highlighting in time with the voice, and you can cut in mid-sentence to redirect it. The mic pauses on its own when the assistant asks a question or shows an approval card, then resumes once you answer. A dropped connection reconnects into the same conversation. Voice and typed messages share one thread and one history. The web version already did this. Mobile reaching parity is the part that counts, because the phone is where you are between appointments.

On September 28 invoices gained an AI layout builder. Describe the invoice you want and it builds the layout. Upload one you already send and it recreates it. From there you adjust individual elements in a visual builder, step back to an earlier version from the chat, then publish the layout to the invoices your customers see. Layouts sit in their own list with preview, clone, rename and delete, and each one keeps a record of who changed it. Invoice design in HighLevel is no longer a choice between fixed templates.

Profile and business cards now build themselves, and QR analytics got specific

On October 27 a QR icon appeared in the sub-account top bar, and behind it sit two cards that were already built. The Profile Card comes from your user details and belongs to you, so every team member gets their own. The Business Card comes from the sub-account's business details and is shared by everyone in it. Both are generated in the background the first time the icon is clicked, so there is nothing to fill in.

  • Links, a QR code and a wallet pass. Each card takes X, Instagram, Facebook, LinkedIn and YouTube links plus custom ones for a website, email, booking page or portfolio, and has its own link and downloadable QR code you can print for offline marketing or save to a phone wallet
  • Exchange Contact turns a scan into a record. Save Contact is on by default, so a scan downloads your details to the other person's phone. Exchange Contact runs the other way and is off until you enable it in the card editor: the person scanning sends you their details, a contact is created in the CRM and the card it came from is recorded. It uses the built-in form or any HighLevel form you point it at
  • System folders on the QR list. Profile Cards and Business Cards appear as managed folders on the QR Codes page. They cannot be renamed or deleted and new codes cannot be created inside them, but the cards themselves stay fully editable, design, colours and landing page included
  • Device and browser breakdown. QR analytics used to stop at total and unique scans. Scans now break down by device type and by browser, so a code scanned almost entirely on phones tells you which version of the destination page is the one worth working on
  • Geography down to the city. Scans report by country, state or region and city, which is the difference between knowing a flyer run worked and knowing which town it worked in

Each card tracks its own scans, so a team of ten has ten separate numbers rather than one pooled count. The capture path is the part worth setting up deliberately, because an event scan that lands as a contact in your CRM is the only version of this that produces a follow-up.

Sub-account features, dashboard widgets and portable segments

On October 14 feature access became a bulk action. Select the sub-accounts you want on the Sub-Accounts page, click Update features, pick the features to change, then set each one to Enable or Disable and click Apply changes. Features can be searched, selected in whole groups, or switched with Enable all and Disable all, and any feature can be dropped from the list before you commit. The changes are queued, take effect in about two to five minutes, and every request is traceable in Bulk Actions History. Rolling a new feature out across a portfolio, or restricting one, stops being a sub-account-by-sub-account job.

On September 24 general widgets on sub-account dashboards became editable, and on September 29 segments became portable. Taken together these are the operational changes agencies will feel first.

  • Permanent filter conditions. A Conditions tab filters any general widget by assigned user, tags, source, lead value, custom fields and attribution, and the filter holds every time the dashboard is opened rather than resetting
  • Date range overrides. Lock a single widget to its own period, independent of the dashboard date filter, and compare it against last period or last year. Each widget face now shows which date property and range it calculates on
  • Smart Lists travel in snapshots. Company Smart Lists and Custom Object Smart Lists became snapshot assets on September 29. Configure segmentation once and deploy it to every destination sub-account. Refreshing an existing snapshot picks up lists created since it was saved. Include the objects and fields the list filters on in the same snapshot, otherwise it arrives with nothing to filter on
  • Field options can finally be deleted. Dropdown, multi-select, radio and checkbox options on Custom Objects and Companies can be removed, matching what Contact and Opportunity fields already allowed. Records holding a deleted option keep the value, the option simply leaves future selection, and the interface shows how many records use it before you commit
  • New custom field types travel too. On October 10 snapshot and public API support went live for the custom field types released in September: URL, Date and Time, Rich text, Users and image options in Radio and Checkbox fields. Those field configurations now replicate across sub-accounts inside a snapshot instead of being rebuilt by hand, and the create custom field endpoint accepts every new type, so an integration can create and update them programmatically. Both are live in production for all accounts, and field types already switched on through Labs need no further activation

Snapshot level segmentation is the piece to test first. White label agency setups rebuild the same Smart Lists in every client account, and this removes that step from onboarding. With the new field types in snapshots as well, a destination sub-account can arrive with its objects, its fields and its segments already built. The CRM guide covers how companies, custom objects and smart lists fit together.

Ad Manager, Communities, gift cards and a Mexico number warning

On September 29 Ad Manager gained Meta campaign budget control. Choose whether the budget sits at campaign level, where Meta distributes it across ad sets automatically, or at ad set level where you set each one yourself. Budget sharing across ad sets can be switched on, and supported budget and bid strategy settings are managed inside the campaign builder instead of in Meta.

On October 8 Community posting opened up to the API. The Social Planner public API can now list eligible community members, check that a member is still active and permitted, and schedule or publish a community post as a specific member. Social Planner rechecks eligibility at publish time, so a post never goes out under the wrong identity, it fails with a readable error instead. Connect the group first under Marketing, Social Planner, Settings, Social Accounts. Draft posts need no member selection. This matters if you run announcements for client communities from outside the platform, which until now meant posting by hand.

On September 30 Community email unsubscribes stopped being a platform wide switch. Clicking unsubscribe in a Communities email could previously opt a member out of all platform email. The preference now applies only to the community that sent the email and the specific notification type, such as a comment mention or an at-everyone post. If you run a community alongside marketing email to the same people, this closes a quiet list attrition path you probably could not see.

On October 15 gift card communications started speaking the customer's language. Purchase confirmations, recipient emails and SMS notifications are translated from the selected locale, and the previews in the gift card send flow reflect the locale across every delivery method, with email previews showing translated content for both sample and real card data. Translation is automatic from the locale, so there is nothing to configure and nothing to maintain per language.

Mexico local numbers no longer send SMS. An upstream carrier migration ended SMS support on Mexico local phone numbers, and on September 29 HighLevel added an SMS unavailable pill in the Phone Numbers table under Manage Numbers so admins can spot affected numbers at a glance. It applies to existing and newly purchased numbers. Audit your Mexico inventory now. The voice minute increase announced for October 13 was cancelled on October 12, but the September 21 notice covering international SMS rates and monthly number rentals still takes effect that day.

Twenty-two days in, October has produced a withdrawn price rise, a storefront, a backend, a GitHub sync, a native chat widget, a rebuilt WhatsApp template builder, a site-wide schema dashboard, one better report, automated internal comments, bulk feature control across sub-accounts and three fixes that remove support tickets. The September releases described above are documented in full in the September 2026 roundup, and this page fills out as October does. October 13 has passed, and it moved international SMS rates and monthly phone number rentals only, because the voice increase was withdrawn the day before it was due. The weekly roundup tracks releases as they land.

More HighLevel updates

Ready to try it yourself?

October 2026 HighLevel releases. October 30 added Hosted Numbers for SMS in Labs, SaaS usage limits for Voice AI and Conversation AI, and mobile call quality, every call recording and Answered By. October 29 brought the Router and Loop workflow actions, AI video generation and video analysis, skills and web search in the AI Agent action, client portal service booking and a calendar view settings panel. October 27 and 28 brought outbound Conversation AI, Workflow Maps, workflows in MCP and a REST API, Client App Builder, Managed Agents with subagents, rental deposits and RCS in Labs. Earlier: the MM Lite deadline for WhatsApp marketing and the cancelled voice price rise.

Try HighLevel Free for 30 Days →

Frequently Asked Questions

Can you use an existing phone number for SMS in HighLevel without porting it?

Yes, through Hosted Numbers, released in Labs on October 30, 2026. An eligible US local or toll free number can be hosted for SMS in HighLevel while voice calls continue through the number's current carrier, so the business keeps the number its customers already know. Agency owners and admins start the request from Settings, Phone System, Manage Numbers, Add Number, Host a Number. The guided flow checks eligibility, verifies ownership with an automated call, takes the service address and sends a letter of authorization for electronic signature. The Labs release covers eligible US local numbers, SMS hosting and one number per request.

What is the difference between a hosted number and an SMS ready number in HighLevel?

Hosted means the hosting process finished. SMS ready means the number can actually send, which also requires location identity verification where it applies, location-level A2P registration and messaging sender setup. If the sub-account already has an applicable approved A2P registration, HighLevel reuses it, so there is no need to register again or pick a campaign for every hosted number. Toll free numbers also need toll free verification. A number can show as Hosted and still not send until the readiness steps are complete.

Can you set usage limits for HighLevel SaaS clients?

Yes. Usage limits went live for SaaS clients on October 30, 2026, as a Labs beta for Pro agencies, covering Voice AI and Conversation AI. You set a limit per client and choose what happens when it is reached, bill for the extra usage or stop the action. Clients see their usage against the limit on their billing page with the reset date, and Manage Client lists every client with a limit filtered by product or by how much is used. A client who cannot upgrade their plan can request a higher limit, and you approve it or reject it with a reason. Limits for the phone system, email and automations, plus alerts at 50, 80 and 100 percent, are listed as coming.

Does the HighLevel mobile app show who answered a call?

Yes, from version 4.27 on iOS and Android. Call details now include Answered By, showing the teammate who picked up, or the phone number if no teammate answered, plus an icon for the device used, web app, mobile app, phone number or SIP. It works on regular, IVR and WhatsApp calls. Open Conversations, long-press the call message and choose Details. Numbers are masked when a call was answered on a phone number, and no icon appears when the device is unknown.

Why does the HighLevel mobile app only play one recording for a transferred call?

It no longer does. From version 4.27 the mobile app shows every recording on a call rather than just the first, each playing on its own and each with its own transcript. Look for the 1/2 and 2/2 labels in Recents, a contact's Call Details or a call in Conversations. Single-recording calls look the same as before, and a transcript appears only when the backend has one for that recording.

Is HighLevel WhatsApp marketing moving to MM Lite?

Yes. Meta has confirmed that marketing messages stop being delivered through the Cloud API on October 13, 2027, and HighLevel published the notice on October 19, 2026. From that date every WhatsApp marketing send has to route through Meta's Marketing Messages API, which HighLevel supports as MM Lite. Service and utility messages are not affected and continue on the Cloud API. Activation is per sub-account under Settings, then WhatsApp, where an activation banner appears, and it takes a few minutes. Existing phone numbers and approved marketing templates carry over with no re-creation and no re-approval, and campaigns keep running through activation.

What does MM Lite do that the WhatsApp Cloud API cannot?

Three things. Delivery decisions are informed by engagement signals, and Meta's A/B test across more than 12 million messages measured up to 9 percent higher delivery for high-engagement messages on the same send volume. Creative formats include animated GIF template headers, Android app deep links and a pilot that enhances template images automatically, none of which the Cloud API supports. Reporting covers the funnel from message through to purchase, plus spend per template, cost per click and cost per delivery, instead of stopping at delivered and read. Meta has said the Cloud API stays available for service and utility traffic but receives no new marketing capability.

Is HighLevel voice pricing changing on October 13, 2026?

No. The voice pricing update announced on October 1, 2026 was cancelled on October 12, one day before it was due to take effect. HighLevel says the underlying providers confirmed the cancellation, so the affected voice rates remain unchanged and no action is required. The withdrawn notice would have moved local incoming minutes to $0.0115 per minute in fifteen countries, mobile incoming minutes to $0.0125 in seven, the United Arab Emirates toll free incoming rate from $0.3675 to $0.4220, and outbound rates for 21 destinations including Morocco mobile from $1.0748 to $1.2354 per minute. None of that applies. The messaging and phone number changes announced on September 21, 2026 are a separate notice and do still take effect on October 13, 2026, including Hong Kong local numbers going from $6.00 to $14.70 per month.

Why do Mexico local numbers show SMS unavailable in HighLevel?

Because an upstream carrier migration ended SMS support on Mexico local phone numbers. On September 29, 2026 HighLevel added an SMS unavailable status pill to those numbers in the Phone Numbers table under Manage Numbers, so admins can see which numbers lost the capability without testing them. It applies to existing numbers and to newly purchased ones. Voice on those numbers is unaffected, and the Mexico inbound and outbound minute increases announced for October 13, 2026 were cancelled on October 12, 2026.

Can HighLevel dashboard widgets have permanent filters?

Yes, since September 24, 2026. General widgets on a sub-account dashboard are editable from Edit dashboard, then the three-dot menu on the widget. The Conditions tab adds filters by assigned user, tags, source, lead value, custom fields and attribution, and those conditions hold every time the dashboard is opened. The Configure tab renames the widget and sets a date range override independent of the dashboard-level date filter. Widgets can be duplicated, themed and set to compare the current period against last period or last year. Available fields vary by widget type, and Funnel and Manual Actions widgets are not configurable yet.

What changed in the HighLevel Lead Source Report in October 2026?

Five things, released on October 6, 2026. Every source row now shows a monetary value for Open, Won, Lost and Abandoned instead of a value on the total only. Any count greater than zero is clickable and opens the exact opportunities behind it. The Source filter accepts Is empty and Is not empty. Opportunities with no source appear in an N/A row rather than being excluded from the report. Row heights and column widths were also tidied for smaller screens. Values are visible only to users with permission to view lead value, and HighLevel lists support for a custom field as the revenue source as the next step.

How do I check A2P registration status for a HighLevel phone number?

Open Manage Numbers, then Phone Numbers. Since October 6, 2026 every number linked to an A2P campaign shows a live carrier registration status of Pending, Registered or Failed, and a number fully registered with carriers also carries an A2P Verified tag. Existing numbers were updated automatically, so there is nothing to switch on. Registration usually completes within 24 hours but the timing belongs to the carrier, so Pending inside that window is expected rather than broken. Waiting for Registered before the first send is what prevents error 30034, the failure that happens when a message goes out from a number carriers have not approved yet.

Can AI Studio apps take payments in HighLevel?

Yes, since October 8, 2026. Open the project, then More, then Payments, switch payments on, choose test or live mode, and select which catalog products the app can sell. Pricing is one fixed price, a chosen set of variations, or the product's full catalog range, and only what you allow can be shown or charged. Ask the agent to add payments and it builds a storefront and checkout, with the payment code managed by the platform so edits to it are restored instead of breaking the cart. Checkout is hosted by HighLevel and supports PayPal express, coupon codes, subscriptions and one-click post-purchase upsells on saved cards from Stripe with 3D Secure, Square, Authorize.net, NMI, Adyen and Mercado Pago. It uses the products, orders, payment providers and subscriptions already in your account, so there is nothing extra to set up. One limit: payments works only on apps built on TanStack Start, so an existing React project needs Project Settings, then Migrate to SSR, which creates an upgraded copy and leaves the original live.

Do HighLevel AI Studio sites have built-in analytics?

Yes, since October 8, 2026, with no tracking code to install. Under the project, More, then Analytics, the Overview shows page views, unique visitors, opt-ins and opt-in rate with trends and the change on the previous period, plus top pages, time on page, traffic sources, devices, browsers and countries, regions and cities. Form Submissions lists every lead the site's forms collected with a link to the contact, filterable by form and date, and a workflow can be created from there. Sites, then Analytics, then AI Studio compares every project alongside funnels and websites. Contacts created from a site keep their visit history under Contact details, Activity. Forms connected before October 8 keep counting opt-ins, but tying a submission to the pages that visitor viewed first needs AI Studio to reconnect form tracking and the site published again.

What is AI Studio Cloud in HighLevel?

A private backend for a single AI Studio project, released on October 9, 2026. Open the project, then More, then Cloud, and switch it on from Overview. In about a minute the project has a database, user accounts and file storage with nothing to connect, and you can keep building while it sets up. The database view lists every table with its approximate row count and whether row-level security is on, rows and cells are editable in place, and asking chat for a new table or policy writes the schema. Daily Backups snapshot schema and rows but not storage files or app code, and Restore swaps the live database for a snapshot while database access pauses for a few minutes. Users covers signups, invitations, sign-in detail and deletion, with email confirmation required by default, optional rejection of passwords found in public breaches and your own password rules, although version one supports email and password sign-in only. Storage keeps files in public or private buckets with an upload size limit and RLS policies. Pricing is hourly. Database size starts at Tiny, which is free, and moves between Mini, Small, Medium, Large and X-Large once an hour, database storage grows but never shrinks and bills for reserved space even when paused, and file storage bills for what is stored. Cloud needs an app built on TanStack Start, so a React project goes through Project Settings, then Migrate to SSR first, and migrating does not switch Cloud on. Remove Cloud permanently deletes that project's database, users and files, so export tables as CSV and download the files first.

Can the new HighLevel custom field types be used in snapshots and the public API?

Yes, since October 10, 2026. The custom field types released in September, URL, Date and Time, Rich text, Users and image support in Radio and Checkbox fields, can be included in snapshots, so field configurations replicate to destination sub-accounts instead of being recreated by hand in each one. The public API supports them as well, including the create custom field endpoint, so an integration can create and update these fields programmatically and keep their data in sync with outside systems. Both are live in production for all accounts, and field types already enabled through Labs need no extra activation.

What is the SEO and AEO tab in HighLevel websites and funnels?

A schema markup dashboard released on October 12, 2026 at website and funnel level, so schema no longer has to be checked one page at a time in the builder. Go to Sites, then Websites or Funnels, select one, then SEO and AEO. A Schema coverage card rates the site Needs work, Getting there, Good or Excellent, with counts for Up to date, Might be outdated and No schema. Each page shows its schema types, its status and when the schema was last updated, and the list can be searched by page name or path and filtered by status. Select pages and click Generate schema to have AI write it in bulk, or use Add schema and the tag editor to work by hand. Generated or edited schema stays Unsaved until you click Save and publish, and it can be discarded. Pages edited and republished two or more times since their schema was created are flagged Might be outdated. Existing schema appears automatically, there is no data loss, schema editing in the page builder still works and both views stay in sync. One caution: running Generate schema on a page that already has schema replaces it entirely.

When does the HighLevel Meta Pixel InitiateCheckout event fire now?

Since October 12, 2026 it fires when a buyer reaches checkout rather than when they submit payment. On store checkout pages, and on any page where InitiateCheckout is set at page level, it fires on page load. Everywhere else it fires on the buyer's first click, tap or keystroke in an order form, covering one-step, two-step, popup and the new order form. It fires at most once per page visit even where there are several order forms or triggers, and autofill, password managers and automatically focused fields do not trigger it. A buyer returning through an abandoned cart link triggers it once the saved cart has loaded, so the value and products are included, and a buyer going straight to an express wallet such as Apple Pay triggers it when payment starts. AddPaymentInfo now also sends order value, currency, product IDs and hashed email, phone, name and external ID where available. Expect InitiateCheckout counts to rise, because the event now includes buyers who reach checkout and leave without paying, so rebaseline any report or automated rule using cost per InitiateCheckout. Setup, cookie consent handling and Conversions API deduplication are unchanged, page-level settings can take up to 24 hours on pages already published unless you edit and save the page, and nothing changes if InitiateCheckout is not enabled on your pixel.

Can HighLevel workflows add internal comments to a conversation?

Yes, since October 14, 2026. A new Add Internal Comment action is available in the workflow builder under Automation, then Workflows. Add the action to a new or existing workflow, write the comment, @mention the users who should be notified, then save and publish. When the workflow runs the comment is posted into the conversation history and is visible to your team only, so it is never delivered to the contact. Typical uses are a handover note on a high priority lead, next step instructions attached on a stage change and account context that until now sat in a separate chat tool.

Can AI Studio apps have a live chat widget?

Yes, since October 14, 2026. Ask AI Studio to add a live chat and it creates a real HighLevel chat widget, names it after the project and builds it into the app's own code, so it uses your site's colors, fonts and layout. Two types are available: Live Chat, and Email Chat for visitors who leave a message and get a reply by email. A project can run more than one widget. Underneath it is the standard HighLevel widget, so visitor sessions use secure tokens that never reach the app's code, a captcha appears when a bot is suspected, your spam settings and rate limits apply, attachments and conversation history work, and business hours, inactivity timeouts, chat ratings and chat-ended messages follow your widget settings. Every message lands in HighLevel Conversations. More, then Analytics, then Chat Widgets reports chat loads, opens, open rate and satisfaction plus a lead table with the page each chat started on, and More, then Chat Widgets manages the widget without leaving AI Studio. Live Chat and Email Chat work in both TanStack Start and React projects, so unlike payments and Cloud this needs no migration. Voice chat, WhatsApp, Facebook Messenger and Instagram widgets are not supported in AI Studio yet.

Can you enable a feature for several HighLevel sub-accounts at once?

Yes, since October 14, 2026. On the Sub-Accounts listing page, select the sub-accounts you want and click Update features. Choose the features to change and click Next, then set each one to Enable or Disable and click Apply changes. You can search for features, select whole groups at once, use Enable all or Disable all, and remove a feature from the list before applying. The changes are queued and processed for you, take effect in about two to five minutes, and the status of every request is visible in Bulk Actions History. This replaces opening each sub-account to change its features one at a time.

Can WhatsApp template media change per contact in HighLevel?

Yes, since October 15, 2026. A media header field in the WhatsApp template builder accepts custom fields and dynamic values instead of only one static public URL. At send time the value resolves for that specific contact and Meta receives a concrete URL for that message, so a single approved template can deliver each recipient their own invoice, ticket, document or image. Before this, a different asset meant a different template and every template went to Meta for approval. Dynamic media headers are supported in workflows and automations, conversations, bulk actions, calendar appointments, reputation review requests and missed call text back. Paired with an image personalisation service, one bulk send can render a unique image per recipient. The same release rebuilt the create and edit experience into a single layout with a live phone preview, used by AI template generation and by the template builder opened from Reputation.

What is a WhatsApp limited time offer template in HighLevel?

A marketing template type released on October 15, 2026 that carries a countdown clock WhatsApp renders inside the customer's chat. Create it under Settings, WhatsApp, Templates, Create template, set Category to Marketing and Type to Limited time offer, then add an offer heading, image header, body, discount code and URL and submit for Meta approval as normal. The clock ticks down to the expiry you choose and turns red in the final hour, so the deadline is drawn by WhatsApp rather than written into your copy or baked into an image. The discount code sits under the clock and copies with one tap, with a URL button beside it, and Meta requires both buttons so neither can be removed. The expiry is set per send, not per template, so one approved template runs campaign after campaign with no new approval. It sends from conversations, bulk actions, workflows and calendar sends, and the AI template writer can draft one. Meta still accepts a send carrying an expiry that has already passed, so check the date on each campaign.

Can AI Studio projects sync with GitHub?

Yes, since October 15, 2026. In a project go to More, then Git, and connect a personal GitHub account or an organization, shared with the sub-account or kept to yourself. A private repository named after the project is created with the first push already done, and reconnecting reuses the old repo rather than force-pushing. Sync runs both ways automatically: every saved version is pushed, whether it came from a prompt, a code edit, a visual edit or a restore, and new commits on the connected branch are pulled back, with a fresh preview built in the background behind an Update preview button. Either direction can be switched off, and Push now and Pull now are always available. It starts on main and the branch can be changed, only the connected branch syncs, and diverged branches offer Force push or New branch. Pushes are scanned for AWS, GitHub, Stripe and OpenAI keys and private keys and blocked on a match. It works on React projects built with Vite and on TanStack Start, so no migration is needed, and commits are authored under the platform's own identity rather than yours. A new Manage Git and download code permission is on by default for every role and can be switched off per user under Settings, Team, Permissions, AI Studio.

Why did HighLevel call tracking show Facebook ad calls as Direct?

Because until October 16, 2026 number pool call tracking could not read most paid Facebook traffic, so those calls were filed as Facebook or Direct and paid calls could not be separated from organic ones. Since October 16 they are labeled Facebook Paid, and calls from TikTok, LinkedIn, Microsoft Bing ads and Google ads clicked inside apps carry their own platform labels. A link tagged with utm_medium=cpc resolves to the platform plus Paid, such as Bing Paid, and where a link carries parameters from several platforms the first match wins, checked in the order Google, Facebook, TikTok, LinkedIn. Visits from ChatGPT and comparable assistants are reported as their own source. There is nothing to switch on because the number pool tracking script updates itself, and the improved labels show in call reports and call details. Two limits are worth knowing. The change applies to new calls only, so past calls keep the source they already have, and a visitor whose browser hides the referrer on a link with no ad tracking parameters still shows as Direct.

Can a HighLevel number pool ring more than one user?

Yes, since October 16, 2026. Open the number pool settings and turn on Ring Incoming Call to selected Users, then pick up to six team members. Every selected person's devices ring at the same time, each using the ring channel that person has configured, and the selection is saved for the next time the pool is opened. Each user's device preference is set under Settings, then Users or My Staff, then Edit, then Call and voicemail, in Default channel for ring all, so one person can take the call on the web app while another takes it on mobile, a desk phone or their own phone number. Before this release, number pool calls could not be routed to specific team members' devices.

Can you filter which HubSpot records import into HighLevel?

Yes, since October 22, 2026. Go to Contacts, Companies or Opportunities, click Import and choose Import from HubSpot, then continue through the wizard to Preview and import. The pencil icon on any records row opens a filter for that object, and Contacts, Deals, Companies and Tasks are filtered separately. The fields available are HubSpot Created Date and HubSpot Last Modified Date. Conditions cover relative ranges such as In the last 6 months and This quarter, and exact ones with On, Before, After and Between, and the exact date range a filter covers is shown before it is applied. Add nested filter combines conditions with AND or OR, and Apply filters to all objects reuses the same conditions across every selected object. Record counts update immediately, so the number found and the number that will import are both visible before Confirm and start import. Filters are optional and leaving them empty imports everything. One detail to watch: In the last excludes today.

Can you attach a form to a HighLevel booking you create yourself?

Yes, since October 22, 2026, for service and rental bookings created from the in-app appointment modal. Select the contact, then click Attach form in the Booking form section and complete the form on the customer's behalf before creating the booking. The form configured for that service or rental is preselected, and any other available form can be chosen from the dropdown. First name, last name, email and phone are prefilled from the selected contact, and CAPTCHA, product selling and payment elements are hidden because a team member is filling the form rather than a customer. The form is optional, so bookings can still be created without one. The restriction is timing: a form can be attached only while the booking is being created, and it cannot be added or edited afterwards. The submitted form can be viewed any time from the appointment details.

Can HighLevel workflows be triggered by Social Planner posts?

Yes, since October 25, 2026. A Social Planner Posts trigger in Workflows fires on six post statuses: Approval Requested, Approved, Approval Rejected, Scheduled, Published and Failed. Add it under Automation, then Workflows, and filter by category, platform, post status, post type, social account and post tag. Platforms covered are Facebook, Instagram, Bluesky, Google Business Profile, LinkedIn, Pinterest, Threads, TikTok and YouTube, and post types include Carousels, Posts, Reels, Shorts and Stories. The common use is an internal notification on Failed, so a post that did not publish is caught the same day, alongside automated handling of approval requests. Only social accounts connected through Social Planner are supported, accounts connected through Integrations do not fire the trigger yet.

Can HighLevel Reviews AI responses be kept on brand?

Yes, since October 26, 2026. Each Reviews AI agent takes brand voice settings: a response style of we or I, the business name, tone traits, a response length and character limit and a sign-off or footer preference. Quality guardrails sit alongside them, so an agent can remember its recent responses, avoid repeating the same opening line and refuse risky language such as refund promises, guarantees, admissions of fault, liability statements and medical or financial advice. Sensitive reviews can be held for human review instead of being published automatically. A live response preview shows how the agent would answer a positive, a neutral and a negative review, and the samples can be regenerated while tone and prompts are still being adjusted, so the agent is tested before it goes live.

Can an agency manage Reviews AI across several HighLevel sub-accounts?

Yes, since October 26, 2026. Go to Agency, then AI Suite, then AI Setup, select one sub-account or several, and set the response mode to Off, Suggestive or Auto responses. The update applies in bulk rather than account by account, and the two views stay in sync, so a sub-account changing its own mode under Settings, then Reviews AI is reflected back in the agency view. Reviews AI can also be disabled for a sub-account from Agency, then Sub-Accounts, then Features and Limits. Disabling it hides the Reviews AI settings, AI Reply and Regenerate for that sub-account and blocks every AI review reply, including manual AI replies, auto responses and drip campaign replies.

What is HighLevel Social Intelligence?

A social listening and brand monitoring workspace released in limited beta on October 26, 2026, at Marketing, then Social Planner, then Social Intelligence. Overview gives a snapshot of brand mentions, industry trends, share of conversation against up to three competitors, post ideas and an AI summary. Explore searches a brand, keyword, hashtag or topic for mention volume, sentiment, engagement patterns and where and when conversations are happening. Saving a search as a topic unlocks 7, 15 or 30 days of historical insight and ongoing monitoring, and topics can be paused or resumed. Alerts cover mentions, sentiment, engagement, impressions and share of voice with adjustable sensitivity and recipients, sent daily or weekly by email. The existing Social Listening experience now sits under a Trends tab in the same workspace. Limits during the beta: it is open to selected users only, each location gets 100 searches for the whole beta period, up to 50 topics can be monitored at once with paused topics excluded, each topic supports three alerts, Overview looks back 7 days, and the dedicated Competitors tab is listed as coming.

How do clients connect their own social accounts to a HighLevel sub-account?

Through Social Connect Invite, released on October 26, 2026. Enable it under Settings, then Labs, then go to Marketing, then Social Planner, then Settings, then Social Accounts, or use the plus Socials button, and generate an invite with an auto-generated or custom password. Send it from Social Planner or copy the link and password and share them however you prefer. The client connects their accounts on a secure external page, and the accounts are added to the correct sub-account automatically, so no workspace access is granted and no credentials change hands. Invites are managed at Settings, then Manage Links, then Social Connect Link, where an invite can be extended, regenerated or revoked, and revoking takes effect immediately. Links expire 30 days after creation. Anyone holding both the link and the password can connect accounts to your sub-account, so share the pair only with the client it was made for.

Do social comments appear in HighLevel Conversations?

Yes, since October 26, 2026. Comments on posts published through Social Planner sync into the Conversations team inbox alongside SMS, email and DMs, with the commenter name or handle, a comment preview, the platform, a timestamp and the original post with a View Post link, and a Social Comments filter to isolate them. Every new commenter is created as a contact, and a repeat commenter is matched to the existing record rather than duplicated. A reply can be posted publicly under the comment or sent as a private DM in a standard inbox thread. Buying-intent keywords such as price, interested or how much, and comments containing questions, can be prioritized. Agency admins control comment sync at Agency Settings, then Social Comments, and sub-accounts choose which platforms sync at Marketing, then Social Planner, then Settings, then Comments. Facebook, Instagram and TikTok are supported at launch, with LinkedIn, YouTube, Pinterest, Threads, Bluesky and Community listed as coming. Only comments on posts published through Social Planner are synced, and turning sync off stops new comments without deleting existing contacts, conversations or workflows.

Can you search messages in HighLevel Conversations?

Yes, since October 26, 2026, at two levels. Cross-conversation search is the search icon at the top of the conversation list: type any word or phrase and results appear as you type, with a total count and a preview of the matching message for each contact, keyword highlighted, and clicking a result opens that conversation at the match. The same box still accepts a contact name, email address or phone number. In-conversation search is the search icon in the conversation header: every match is highlighted, a counter shows your position such as 6 of 6, and up and down arrows jump between matches, with the keyword prefilled if you arrived from an inbox-wide search. Results can be narrowed by date range, channel across email, SMS, Facebook, WhatsApp and Instagram, direction, owner, has attachments and followers. HighLevel lists keyword search as one of the most requested items on its Ideas Board.

What new filters were added to HighLevel Conversations in October 2026?

Six, released on October 26, 2026 in the Conversations filter panel. Opportunity Pipeline filters by the pipeline attached to a contact opportunity, and a contact with opportunities across several pipelines appears if at least one matches. Opportunity Stage filters by stage, with the available stages shown for the selected pipeline. Opportunity Status filters by open, won, lost or abandoned. Channel Account filters by the specific channel the most recent message used, such as one phone number, one connected Facebook page or one Instagram account. Last Call Status filters by the outcome of the most recent call, including completed, no answer and voicemail. Contact Type filters by Leads, Customers and any custom contact type configured in the sub-account. Open Conversations, click the Filters icon, pick a filter type from the dropdown, select one or more values and click Apply, and add more filters to combine conditions. All six work with the existing filters using AND and OR conditions, and Clear resets the inbox.

What is a WhatsApp carousel template in HighLevel?

A marketing template type released on October 26, 2026 that carries between 2 and 10 swipeable cards in a single message. Create it under Settings, WhatsApp, Templates, Create template, set Category to Marketing and choose the Carousel type, write the message body once, then add each card with its own image or video header, optional text and up to two buttons, and submit for Meta approval. WhatsApp bills a carousel as one marketing message and counts it once against your per-customer frequency limits rather than ten times. Buttons mix quick reply, call and website types, a quick reply answers in the thread and fires a webhook you can route into a workflow, and each card's website button can carry its own variable so card one and card two go to different pages. Two rules to plan for: card count is locked at approval, so a five-card template always sends five and a four-card version is a second template, and every card has to match the others, meaning if one card has body text they all need it and if one has two buttons they all need two. It sends from Conversations, workflows and bulk actions.

Can a HighLevel form be filled in inside WhatsApp?

Yes, since October 26, 2026. Go to Settings, then WhatsApp, then Flows, click Create Flow and choose Form submission, select a form you have already built, check the live preview, configure the WhatsApp template that carries it including the button customers tap, then publish and send it from Conversations, a workflow or a bulk action. The customer answers the questions in the chat window with no link to tap and no browser to open, which removes the redirect where leads usually go quiet. Responses are stored in Forms, then Submissions and mapped to the right contact automatically, so nothing is copied out of a chat by hand. Field types WhatsApp Flows cannot render are left out before publishing rather than discovered from a broken send, and once a flow is published later edits to the underlying form do not change what it collects, so data stays consistent across a campaign.

Can WhatsApp calls be recorded in HighLevel?

Yes, since October 26, 2026. Recording is enabled once for a number and every WhatsApp call on that number is captured from then on. Finished calls appear in Conversations as a playable recording alongside that customer's messages, with play, scrub and speed controls, and a transcript beside the audio separating what your team said from what the customer said. The language of the call is detected automatically, so transcripts work across markets. Consent is built into the flow: before recording starts the caller hears a short announcement explaining that the call is being recorded and why, and you choose both the language it is spoken in, including Spanish, Portuguese, French, German, Italian, Dutch, Hindi, Kannada, Telugu and Vietnamese, and the reason they hear.

Does HighLevel WordPress hosting include antivirus?

Yes, since October 26, 2026, on every WordPress site. Go to Sites, then WordPress, open a site, then Antivirus, then Shield. Shield shows how many items need attention and when the site was last scanned, Scan now runs a fresh scan, Action required lists each problem with why it matters, and fixes apply one at a time or all at once. Ten checks return Passed or Action required, covering core and plugin files compared against their official versions, unused plugins, switched-off protections, HTTPS, the WordPress version, supported server software, hidden error messages and search engine visibility. Items can be snoozed until the next scan, for 7 days or for 30, and the last 30 days of fixes stay under Resolved. A separate Hardening tab has ten one-click protections including blocking application passwords and user enumeration, hiding the WordPress version, masking login errors, disabling the file editor and XML-RPC, security headers, login rate limiting and .htaccess hardening. A Threats column on the sites list shows counts for every site and all sites can be scanned at once. Two cautions: every fix changes the live site, so use the Back up first option it offers, and leave Disable XML-RPC off if the site is managed from the WordPress phone app.

Can readers comment on HighLevel blog posts?

Yes, since October 26, 2026. Go to Sites, then Blogs, open the blog site, open the blog post page in the builder, add the Comments element, style it, then save and publish. Readers can write comments, reply to each other, add emoji, attach one image per comment and like or dislike, with threaded conversations and Load more comments on long discussions. Returning readers sign in with their email and a one-time code with no password, and new readers create an account in the same window. Replies from your team carry a Moderator badge. A Comments tab in Blogs shows total comments, comments this month and comments awaiting approval with trend lines, filters by status or post, search, and reply, approve or delete on any comment, plus a one-by-one review of pending comments. Per blog site you can require approval before anything goes live, send an email for every comment or a daily digest, choose the recipients, and apply the same settings across the sub-account. Two limits: comments are switched on for a whole blog site rather than per post, so removing the element removes them from every post, and comments do not appear in Conversations yet and commenters are not added to Contacts.

What are HighLevel profile and business cards?

Two QR-backed digital cards released on October 27, 2026, generated automatically rather than built by hand. Click the new QR icon in the top navigation bar inside a sub-account and both are already there. The Profile Card is built from your user details and is personal to you, so every team member gets their own with their name, title and contact details. The Business Card is built from the sub-account's business details and is shared by everyone in that sub-account. Each card takes links for X, Instagram, Facebook, LinkedIn and YouTube plus custom links such as a website, email, booking page or portfolio, and has its own link and downloadable QR code you can print or save to a phone wallet. Save Contact is enabled on every card, so a scan downloads your details. Exchange Contact is off by default and enabled from the card editor: it lets the person scanning send you their details instead, creating a contact in the CRM and recording which card it came from, using the built-in form or an existing HighLevel form. Each card tracks its own scans. The cards live in managed Profile Cards and Business Cards system folders on the QR Codes list page, which cannot be renamed or deleted, although the cards inside stay fully editable.

Can HighLevel Conversation AI start a conversation first?

Yes. Outbound AI agents arrived on October 27, 2026. Inside the agent, under Deploy then Outbound then Triggers, the agent can send the first message when a form is submitted, a Facebook or Instagram lead ad comes in, a call is missed, a tag is added, an appointment is booked or an opportunity changes stage. It writes a personalised opener, follows up when the contact goes quiet and books the appointment, with no workflow to build. AI Campaigns do the same thing in bulk from the Contacts page, writing each message from that contact's past interactions with the business, which is how database reactivation now runs natively.

Can HighLevel Conversation AI run on Claude or Gemini?

Yes. From October 27, 2026 the Model menu in Conversation AI lists Anthropic Claude and Google Gemini models next to the GPT ones: Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 and Claude Haiku 4.5, plus Gemini 3.8 Flash, Gemini 3.7 Flash and Gemini 3.5 Flash Lite. Each model shows a relative cost indicator, so a high volume FAQ agent can run on a cheap fast model while a sales agent runs on a stronger one. The model is set per agent, so different agents in the same sub-account can use different providers.

What is Workflow Maps in HighLevel?

Workflow Maps, released on October 27, 2026, is a visual view of every workflow in a sub-account together with the folders, tags, forms, calendars, pipelines and custom fields that connect them. It is generated from workflows that already exist, so there is no diagram to maintain. It shows chains where a form starts a workflow, that workflow adds a tag and the tag starts another workflow. It also tracks an individual contact's journey across workflows with the runs and steps behind it, flags workflows with execution errors so a broken automation surfaces on its own, opens linked workflows together in the Workflow Canvas, and includes an AI assistant that answers questions about the account and can build or edit workflows. It lives under Automation, then Workflows, then Workflow Maps.

Does HighLevel have built-in video meetings?

Yes. Meeting Room launched on October 27, 2026. It gives each user a permanent personal room link, supports instant meetings without an appointment, and generates a unique link automatically for calendar appointments. Guests join by link with no account. It includes HD video and audio, screen sharing, live chat, co-hosts, host control of participant microphones, cameras and screen sharing, waiting rooms, passcodes, recordings, transcripts and AI summaries. Breakout rooms shipped at the same time, with up to 100 rooms per meeting, automatic or manual assignment, self-selection, host broadcasts, a timer and automatic return to the main meeting.

Can HighLevel workflows be built from an AI assistant?

Yes, in private beta since October 28, 2026. Workflows are now part of the LeadConnector MCP server, so Claude, ChatGPT, Cursor and other connected assistants can search workflows by name, status, folder or date, read a full definition including triggers, steps, branches and settings, build a workflow from a description and save it as a draft, update and publish it or set it back to draft, and delete one after a second confirmation. Execution logs are the part that saves the most time: asking which steps in a workflow failed this week returns each failed step with the contact and the error, instead of a manual read through the logs. Enrollment, completion and failure counts per step are available by hour, day or week. A Workflows API released the same day covers the same operations from your own code, with a validate call that points at the exact failing step before a definition is saved, and a step types endpoint listing every trigger and action available in the sub-account with its fields. Both are private beta, so access is not automatic.

Can you build a mobile app in HighLevel?

Yes. Client App Builder, released on October 28, 2026, generates a branded Android app from a description. Choose a design system and brand board for the logo, colors and images, and AI builds the screens, design and functionality, connected to the courses, communities, files, appointments and member accounts already in the sub-account. Sign-in, sign-up and member onboarding with join codes are built in, device features such as camera and photo access are supported, custom objects can be managed from inside the app, and a live phone preview shows the result before publishing. The builder produces an installable Android APK with build progress and errors visible. Released alongside it, Vibe Coding does the same for the web portal, with up to 10 custom domains and visual editing where you select an element in the preview and describe the change. Custom Objects in Client App Builder lets members submit records that save straight into the CRM, with visibility set to their own records, their company records or read-only access to all, and payments collected on a record through invoices.

Can HighLevel rentals take a deposit instead of the full amount?

Yes, since October 28, 2026. Partial Payments is configured once in Global Settings: set a deposit of 1 to 99 percent of the booking total and say when the balance is due. Checkout shows Due today and Due on the date with a tooltip explaining the automatic charge, and on the due date the balance can be charged to the card used for the deposit. A security deposit collected upfront can be taken at booking or with the final balance. Two related releases landed the same day. Additional fees are mandatory charges such as cleaning, resort, service or damage waiver fees, built once under Calendar settings, then Rentals, then Additional fees, priced flat or as a percentage of the rental subtotal, charged per booking or per booking unit, with tax support, per-listing overrides and a visible line item on the invoice. Add-ons are the optional version, each with a name, description, image, maximum quantity and tax settings, selected by the customer in the booking widget with the total updating live. Both are built as reusable catalogs, so a fee or an extra is created once and attached to as many listings as needed.

Can an agency see A2P registration status across all sub-accounts?

Yes. Agency Trust Center, released on October 28, 2026, puts A2P 10DLC registrations for every sub-account in one dashboard in Agency view. It tracks Fully Approved, In Review and Rejected counts, surfaces the common brand and campaign rejection reasons with a route through to the affected registrations, and compares chat widget against self-serve registration adoption and approval rates. Brand registrations are viewed and managed from the same place, so a compliance problem does not have to be found by opening each location. RCS messaging reached Labs the same day, with the full sender registration completed inside HighLevel rather than on a provider platform: brand identity with business name, logo, description, contact information and website, the countries where RCS will be used, and sender details covering the use case, the message types and how customers give consent. Agencies choose which sub-accounts to onboard from Labs.

What is the Router action in HighLevel workflows?

A branching action released on October 29, 2026. If/Else stops at the first branch a contact matches, while Router checks every branch and runs each match in turn, left to right. Branches can be Custom rules, Always run or Fallback, the order is set by dragging them in the Branches list, and the test output shows each condition as passed or not with the value the contact actually had.

Can a HighLevel workflow repeat actions for every item in a list?

Yes. The Loop action, released October 29, 2026, runs the actions inside it once per item in a list or once per number in a range. The list can be an array from an earlier step such as an Inbound Webhook, and Custom values, then Loop Item, maps the current item's fields into each action. Iterations run sequentially by default, with a Parallel toggle.

Can HighLevel workflows generate and analyze video?

Yes, through two premium AI actions released October 29, 2026. AI Video Generation creates a 4, 6 or 8 second clip with Veo 3.1 at 720p or 1080p, stores it in the Media Library and returns its URL. AI Analyze Video sends a public video URL to Gemini, which watches the picture and listens to the audio, then answers your prompt. Both are billed per execution.

What are skills in the HighLevel AI Agent action?

Reusable procedures, added October 29, 2026, that an agent loads only when a run needs them. Each has a name, a line saying when to use it and a body of instructions, and the agent reads only the name and that line on every run, so unused skills cost nothing. Several can load together. Four ship built in: Humanizer, Write-well, Cold-email and Sms-writer.

Can the HighLevel AI Agent search the web?

Yes. Web search is a Built-in tools toggle added October 29, 2026. The agent searches only when a task needs facts it does not already have, and a URL pasted into the Instructions points it at a specific page. It is billed per search, at 1 cent on OpenAI and Claude models and 1.4 cents on Gemini models.

Can clients book services in the HighLevel client portal?

Yes, since October 29, 2026. Clients browse services, book, reschedule and cancel inside the portal with their details prefilled, and meetings and service bookings appear together under Appointments as Upcoming and Past. Enable it per booking page under Calendars, then Services, then Booking Pages, with Allow booking from client portal.

launchkit.work is part of the NetPartners Marketing network of GoHighLevel guides, published by Zoltan Juhasz. This site is a GoHighLevel platform overview with playbooks, comparisons and implementation guides.

Go deeper on netpartners.marketing: Pricing plans explained · Every feature explained · What's new in GoHighLevel

Zoltan Juhasz
Zoltan Juhasz
Marketing consultant and HighLevel affiliate. Publisher of netpartners.marketing, the independent GoHighLevel guide for agencies and SaaS resellers.
About LinkedIn YouTube Contact

This site is published by Agence Vesta Inc. (NetPartners Marketing), Saint-Lambert, Quebec, and is funded by affiliate commissions from HighLevel Inc. Links to gohighlevel.com carry an affiliate reference; a subscription started through them earns this site a commission at no extra cost to you. Privacy policy · Terms of service