GoHighLevel AI Builder Test Workflow: How to Test Before You Publish

Short answer: AI Builder in GoHighLevel can now test the workflow it just built or edited. After it finishes, its reply shows a Test workflow button; you pick a contact or a test event, click Start test run, and AI Builder runs the steps, marks each one on the canvas and writes a plain summary of what ran, what was skipped and what failed (Official, HighLevel changelog, October 29, 2026). It is a Labs feature: turn on The New Test Workflow in Labs first. Important: the test runs the real steps, so tags are added, notes are written and messages are sent. Test with your own contact.

What the AI Builder test does

AI Builder is the chat panel inside the GoHighLevel workflow editor where you describe an automation in plain words and it builds or changes the steps for you. Until now you had to leave the chat, open the separate test tool and work out what happened yourself. The October 29 changelog entry moves testing into the same conversation.

According to the changelog (Official), once AI Builder has built or edited a workflow, its reply includes a Test workflow button. When you run a test:

• The test runs inside the chat, and the canvas shows the status of each step as it goes.
• AI Builder writes a short summary in plain language: what ran, what was skipped, and any errors or missing data it found.
• You can click any step on the canvas to see its input and output.
• You can change the workflow and run the test again.

Testing stays optional. You can still save and publish without running it. HighLevel has not stated any extra cost for the feature, and the entry does not say when it will leave Labs (not announced yet).

How to set it up, step by step

1. Turn it on in Labs. In your sub-account, open Labs and switch on the feature called The New Test Workflow. Without it, the button does not appear.
2. Open a workflow. Go to Automation, then Workflows, and open the workflow you want to build or change.
3. Ask AI Builder for the change. Describe the workflow or the edit in plain words, for example "when a lead fills out the quote form, tag them, text them within a minute and create an opportunity".
4. Click Test workflow. When AI Builder finishes, the button is in its reply.
5. Choose how to test. Either enroll a contact (pick one) or test using events, then click Start test run.
6. Read the summary. Check which steps ran, which were skipped and why, and any missing data. Click a step on the canvas to see exactly what went in and what came out.
7. Fix, retest, then publish. Ask AI Builder to adjust anything that looks wrong, run the test again, then save and publish.

The warning that matters: it runs real steps

This is not a dry run. The changelog says plainly that a test runs the real steps: tags get added, notes get written and messages get sent. If you enroll a real customer, that customer gets the real text or email. A few rules keep this safe:

• Use your own test contact. Create a contact with your own mobile number and email, and use it for every test. This is what HighLevel recommends.
• Watch for side effects outside GoHighLevel. Webhooks, payment steps and anything that writes to another system may fire. Check those steps before you test, or point them at a test endpoint.
• Clean up afterwards. Remove test tags, test opportunities and test notes so your reports stay accurate.
• Some actions are skipped. The summary and canvas tell you which ones did not run during the test, so do not assume a skipped step works. Check those by hand.

If a workflow has ever reported success while sending the wrong message, read why a GoHighLevel workflow can say it worked and still send the wrong message. The step by step input and output view is the quickest way to catch that kind of mistake before a customer does.

Who should use it

Business owners who build their own automations. If you use AI Builder because you do not want to learn every trigger and action, the plain-language summary is the part you will use most. It tells you, in words, whether the follow-up you asked for actually happens.

Agencies and builders. Testing in the same chat shortens the build, check, fix loop. It pairs well with the other workflow changes from LevelUp week, such as the Router action, which runs every matching branch, and workflow A/B testing. A Router workflow in particular deserves a test run, because one contact can go down several branches.

Teams handing workflows to clients. A tested workflow with a clear summary is easier to hand over. Our guide on stopping clients from breaking workflows after handoff covers the rest of that process.

A quick testing checklist

1. Write down what should happen, in order, before you click Test workflow.
2. Test with your own contact, never a live customer.
3. Compare the AI summary with your list. Any step that is skipped or shows missing data needs a look.
4. Check the messages that arrive on your phone and inbox: right number, right sender name, merge fields filled in.
5. Test the edge cases too: a contact who already has the tag, a contact with no email, a contact who replies.
6. Remove test data, then publish.

If your account already has many workflows and you are not sure which ones still work, a GoHighLevel account audit checks them for you. For ongoing changes and testing every month, see the GHL care plan, or book a free discovery call at meet.yotomations.com.

For every other feature HighLevel shipped during LevelUp week, see our LevelUp 2026 announcements hub.

Yaakov (Koby) Oranski, Founder and CEO of Yotomations

Certified GoHighLevel Admin and Partner. Koby designs CRM migrations, two-way integrations, billing systems and AI agents for service businesses, and writes from the engagements described on this site.

Published 2026-10-30