This GoHighLevel tutorial covers a new AI Agent option inside workflows, and it comes loaded with prebuilt prompts for things like form lead follow-up and no-show recovery. It’s still beta, so you can use it today, but you’ve got to test everything before you trust it with real leads. Here’s exactly where to find it and what it does.
Where to find the GoHighLevel AI Agent
This one’s easy to miss if you’re not looking for it. Inside a workflow, hit the plus sign like you’re adding any other action, then scroll down. You’ll see a new option labeled AI Agent sitting alongside your regular workflow steps.
Here’s the thing — it’s not a separate tool bolted onto GoHighLevel. It lives right inside the workflow builder you already use. That matters, because it means the agent triggers off the same events you’re used to: a form submission, a missed appointment, a tag getting added, whatever starts your workflow already.
How the agent actually works
Once something triggers the workflow — the customer does something, or you do — the agent takes over from there. It’s not running through a fixed set of if-this-then-that steps. It’s working off a prompt you give it, and it decides what to do next based on that prompt and the situation in front of it.
That’s a real shift from how GoHighLevel workflows have worked up to now. A normal workflow is rigid: step one, step two, step three, no matter what. An agent reads the situation first, then acts.
What do the prebuilt AI Agent prompts do?
The prebuilt AI Agent prompts handle common follow-up scenarios automatically — things like reaching out to new form leads, recovering no-shows, nurturing Facebook leads, and nudging stale pipeline deals. Each one is a ready-made starting point you can activate and then adjust to match your own process.
The part that stood out to me is that GoHighLevel didn’t just hand you a blank box and say “good luck.” They shipped a stack of prebuilt prompts you can start from immediately. A few worth knowing:
- Form lead follow-up — the agent picks up a new form submission and starts the follow-up conversation on its own.
- No-show appointment recovery — someone misses their booked call, and the agent works to get them rescheduled.
- Facebook lead nurturing — keeps leads from your Facebook ads warm without you touching the conversation.
- Stale deal nudge agent — pokes at deals that have gone quiet in your pipeline before you lose them entirely.
- Research and enrichment on new contacts — automatically fills in context on a lead the moment they come in.
- Appointment enrichment and confirmation — adds context to new bookings and confirms them.
- Quiz lead scoring and routing — scores leads from a quiz funnel and routes them where they need to go.
- Call transcript summary and action items — turns a call recording into a summary with next steps.
- Task creation from won deals — spins up the follow-on tasks the moment a deal closes.
- Instagram and Facebook comment-to-DM — turns a public comment into a private conversation automatically.
- Lead pipeline tracker — uses memory to track leads through pipeline stages with periodic check-ins.
Nuno’s reaction when he ran through this list says it plainly: “That is amazing, right?” And it is — this is the kind of follow-up most people never get to because they run out of hours in the day, not because they don’t know it matters.
What should you watch out for since it’s beta?
Because it’s beta, treat every prebuilt prompt as an untested draft rather than a finished process. Test with dummy contacts first, check what memory-based agents are actually tracking, read each prompt before activating it, and expect the feature to change without much warning.
Here’s what nobody tells you about beta features: “beta” doesn’t just mean new, it means unfinished. As Nuno puts it, “It’s still beta. So obviously you want to be very smart about how you do this.”
Practically, that means a few things before you turn any of these prompts loose on real leads:
- Test with dummy contacts first. Don’t let an agent’s first conversation ever be with a real prospect.
- Watch what the memory-based agents remember. The lead pipeline tracker relies on memory across check-ins — confirm it’s tracking the right stage before you trust it unsupervised.
- Read the prompt before you activate it. These are prebuilt starting points, not finished scripts for your business. Your tone, your offer, and your follow-up rules aren’t in there by default.
- Expect it to change. Beta features get adjusted, sometimes without much warning. Don’t build a mission-critical process around one until it’s out of beta.
The honest gotcha: a prebuilt prompt handling “no-show recovery” or “stale deal nudge” sounds like it’s ready to run unattended. It’s not. It’s a draft that happens to be a good one. You still own reviewing what it says and how it decides to act, especially early on.
Why this fits how GoHighLevel should be used
This is exactly the kind of feature that goes sideways if you bolt it onto a messy setup. An agent that nudges stale deals is useless if your pipeline stages don’t mean anything, or if it’s one of eight different tools fighting over the same lead. Profitable setups run on a handful of connected tools doing their job well — not thirty half-configured ones. Get the foundation right first, then let the agent work inside it.
If you want to see how workflows, pipelines, and automations fit together as one system instead of scattered pieces, that’s covered start to finish in the GoHighLevel Masterclass.
Frequently asked questions
Where do I find the AI Agent in GoHighLevel?
Open any workflow, click the plus sign to add a new action, then scroll down. AI Agent appears as an option alongside the standard workflow actions.
Is the GoHighLevel AI Agent free to use?
It’s included as part of the workflow builder — there’s no separate purchase for the feature itself, but it’s currently in beta, so functionality and availability can change.
Can I edit the prebuilt AI Agent prompts?
Yes. The prompts for form lead follow-up, no-show recovery, and the others are starting points. You should edit them to match your tone, your offer, and your actual follow-up process before turning them on.
Should I trust the AI Agent with real leads right away?
Not yet. Since it’s beta, test each prompt with dummy contacts first and watch how it behaves before letting it run on live leads unsupervised.
What’s the difference between a regular GoHighLevel workflow and an AI Agent?
A standard workflow follows a fixed sequence of steps every time. An AI Agent reads the context of the trigger and the prompt you gave it, then decides what to do next — it’s reactive to the situation, not just running a script.