GoHighLevel AI Builder Now Configures Workflows with Your Real Forms, Pipelines, Tags and Snippets

Status check: this comes from HighLevel's official changelog entry "AI Builder Now Configures Workflows with Your Account Data", published on 27 October 2026 and tagged New under Automations. The entry does not mention Labs, a beta or a staged rollout, and HighLevel has not presented it as a LevelUp announcement. If you do not see it in your account yet, check again later before you plan client work around it.

When I build workflows for agencies, the slow part is rarely the structure. It was the clicking afterwards: open the trigger, pick the form, open the opportunity step, pick the pipeline, pick the stage, open the email step, find the snippet. AI Builder used to hand you a skeleton and leave all of that to you. This update closes that gap.

What is new

Before, AI Builder created the right triggers and actions, but you still opened each step to choose the form, the pipeline, the stage or the snippet. Now it looks up the records you name and fills them in during the build. HighLevel's own summary is: "Name it in the prompt, and it is already set."

What each step gets set to

The changelog gives these examples:

• Form Submitted: the filter is set to the exact form you named.
• Create Opportunity: the pipeline and stage are set from your account.
• Send Email: the subject and body come from your existing snippet, linked to the action.
• Add Tag: uses the tag that already exists instead of creating a new one.
• Create Task: assigned to the user you named.
• Add to Workflow: points to the workflow you named.
• Update Contact Field: updates the custom field you named.

You can watch it find your data

The build progress view lists each record as AI Builder finds it, for example Found form, Found pipeline and Found email template. I like this a lot. If a line you expected never shows up, you know before you even open the workflow that a name did not match.

Records AI Builder can look up

This is the full list from the changelog:

• Workflows and automation: workflows, campaigns
• CRM and account: users, teams, tags, trigger links, countries, phone numbers, number pools
• Opportunities: pipelines, pipeline stages, opportunity custom fields
• Calendars: calendars, calendar services, calendar service categories, calendar service locations
• Forms and content: forms, surveys, funnels, videos, certificates
• Products and memberships: global products, membership products, membership categories, membership lessons, membership offers, affiliate campaigns
• Fields, objects and values: custom values, contact standard fields, contact custom fields, business custom fields, custom objects, custom object fields
• Templates: email templates, SMS templates, estimate templates, invoice templates
• Social: Facebook pages, Facebook forms, Instagram pages

How to use it, step by step

1. Go to Automation > Workflows.
2. Create a workflow, or open an existing one, and open AI Builder.
3. Write your prompt. Name each record exactly as it appears in your account, for example "Pipeline: Marketing, Stage: New Lead" or "use the existing snippet Welcome Email".
4. Send the prompt and watch the progress view as AI Builder finds each record.
5. Open each step and review the values it set.
6. Test the workflow, then publish.

A worked example prompt

Here is the kind of prompt I would write for a simple quote request flow. Before sending it, I make sure the form, tag, pipeline, stage, snippet and user already exist with these exact names, because AI Builder will not create them.

What I expect back:

• Progress view shows Found form, Found pipeline and the snippet, tag and user being picked up.
• Form Submitted trigger filtered to Free Quote.
• Add Tag using the existing quote-request tag, not a new copy.
• Create Opportunity set to Sales, New Lead.
• Send Email with the subject and body pulled from the Welcome Email snippet.
• Create Task assigned to the named user.

My prompting tips: copy names straight from the account rather than typing them from memory, keep the record type next to the name ("the form Free Quote", "the tag quote-request") so there is no guessing, and say "existing" when you mean an existing record. Spelling and spacing matter: "Free Quote" and "Free Quote Form" are different records.

My review checklist before publishing

HighLevel is clear on this: always review each step before you publish. If a step picked the wrong record, ask AI Builder to change it, or open the action and select the right one. This is the list I run through:

• Trigger: is the Form Submitted filter on the right form, and only that form?
• Tags: does Add Tag use the existing tag, with no new near-duplicate in your tag list?
• Opportunity: right pipeline and right stage? Similar pipeline names are easy to mix up.
• Email: is the snippet linked, and do the subject and body read correctly for this flow?
• Tasks: assigned to the right user?
• Add to Workflow and field updates: pointing to the workflow and custom field you meant?
• Post-Generation To-Do List: settings like credentials still need your decision and show up here. Clear every item.
• Test it: run a test contact through before you publish, and remember a test of real steps can send real messages.

Why this matters for agencies

• Less manual setup: no more setting values one by one after the build.
• Fewer mistakes: HighLevel says that because it uses records that already exist, it does not create duplicate tags or point to the wrong form. The same entry still tells you to review each step, because a step can pick the wrong record.
• Closer to ready: the workflow you get back is much nearer to something you can test and publish.

My take: this makes AI Builder more useful for repeat builds. If you keep the same naming across your sub-accounts (same pipeline names, same snippet names, same tag style), you can reuse one prompt again and again. Consistent naming across your accounts now matters more.

For the rest of today's updates during LevelUp week, see the Founder Plan explained, TikTok Ads and shareable AI agent chat pages and the CRM update with calculated fields and companies.