Home/HighLevel/AI/Managed Agents

HighLevel AI · Managed Agents

GoHighLevel Managed Agents. AI teammates you describe in plain language.

GoHighLevel Managed Agents, formerly called Super Agents, are custom AI teammates you build by describing their job. They fire on real events in your CRM, such as a new contact or a new opportunity, and complete tasks inside HighLevel and in other tools through connectors. They are the open ended layer of HighLevel's AI.

  • Plainlanguage builder
  • CRMevent triggers
  • 1,000runs a month on Unlimited
HighLevel Managed Agent card showing role lead researcher, trigger new opportunity, allowed actions update contact, add note and create task, connectors, and a run log marked run complete

The short answer

HighLevel Managed Agents, formerly called Super Agents, are AI teammates you create in Agent Studio by describing their role and allowed actions in plain language. They run when CRM events happen, such as a contact being created, and act inside HighLevel and in outside tools through connectors. Growth includes 100 runs a month and Unlimited includes 1,000.

Michael is a sponsored HighLevel affiliate. Links to HighLevel use his partner link at no extra cost to you. HighLevel prices and product facts on this page come from HighLevel's own pricing page and help center, checked on the date above.

How it works

Describe the job, pick the trigger.

  1. Open Agent Studio and create a Managed Agent.
  2. Describe the role in plain language: what it does and what it must not do.
  3. Choose the trigger, such as Contact Created, Opportunity Created or Form Submitted.
  4. Allow the actions and connectors it may use.
  5. Test it in the testing panel with realistic requests.
  6. Change its status from Draft to Published and review its runs.

Ideas

Jobs Managed Agents do well.

Lead research

When a contact is created, enrich the record and write a short brief for sales.

Pipeline hygiene

When an opportunity stalls, add a task and draft a check in message.

Handoffs

When a deal is won, create the onboarding tasks and notify the delivery team.

Outside tools

Through connectors, update a project tool or a spreadsheet when something changes in the CRM.

Managed Agents can be cloned and shared through snapshots, which makes them useful for agencies deploying the same agent to many clients.

Which agent

Managed Agents vs Conversation AI and Ask AI.

Managed AgentsConversation AIAsk AI
Works forYour teamYour customersYou
Started byCRM eventsIncoming messagesYour request
Typical jobAutomate tasks across toolsAnswer, qualify, bookCreate, find, act in the app

Worked example

How do you write a job description for a Managed Agent?

A Managed Agent is only as good as the description you give it. Write it the way you would brief a new assistant on day one: the job, the limits and what done looks like.

Here is a description for a lead research agent. When a new contact is created from the website form, read what they submitted. Write a brief of no more than five sentences covering who they are, what they asked for and how urgent it sounds. Save the brief as a note on the contact and create a task for the sales owner, due today. Do not message the contact. Do not change any field other than the note. If the form is empty or looks like spam, add a tag for review and stop.

  • It names the trigger. A new contact from one source, not every contact.
  • It defines the output. A short brief, a note and a task.
  • It sets hard limits. No messages to the customer and no other field changes.
  • It covers the odd case. Spam and empty forms have their own instruction.

The Managed Agents setup guide walks through the builder screens.

Fit

When is a Managed Agent the wrong choice?

  • The job is talking to customers. That is Conversation AI or Voice AI. Managed Agents work for your team.
  • A fixed rule would do. If the logic is always if this then that, a standard workflow is simpler and cheaper.
  • It is one judgment inside an automation. Use a single AI step in a workflow. AI workflows explains the options.
  • The stakes are high and nobody reviews it. Do not give an unattended agent the power to delete records, issue refunds or send messages you have not seen.

The honest test is whether you could describe the job to a person in a paragraph and trust them with it on their first day. If yes, it is a good Managed Agent. What AI can and cannot do covers the wider limits.

Run allowances by plan are in the GoHighLevel AI plans overview. You can build and test a Managed Agent during a 30 day trial through Michael's link.

Avoid

What mistakes do people make with GoHighLevel Managed Agents?

  • Too many actions allowed. Allow only the actions and connectors the job needs.
  • A trigger that fires too often. Every run counts against the allowance. Narrow the trigger.
  • Publishing without testing. Use the testing panel with realistic requests, including bad input.
  • Never reviewing runs. Read the runs weekly for the first month and tighten the description.
  • One agent for five jobs. Give each agent one role. Two small agents are easier to trust than one large one.

Once an agent is proven, agencies can deploy it to other clients through a snapshot. The snapshot guide covers packaging, and which AI plan is best shows which plan includes how many runs.

Michael Reimer
Michael's takeFounder of CRM Pros · HighLevel Commander

Managed Agents are where HighLevel is going. Think of every repetitive task your team does after a CRM event, and write it down as a job description.

Questions

Common questions.

What are GoHighLevel Managed Agents?

They are custom AI teammates that you create in Agent Studio by describing their role and allowed actions in plain language. They run when CRM events happen, such as a contact being created, and act inside HighLevel and in outside tools through connectors. They were formerly called Super Agents.

Were they called Super Agents?

Yes. Managed Agents were formerly called Super Agents. The name changed and the idea stayed the same: an AI teammate you describe in plain language that runs on CRM events. If you see Super Agents in an older tutorial, it means the same feature.

How much do Managed Agents cost?

Managed Agents are billed by runs. AI Employee Growth at $50 a month includes 100 runs, and Unlimited at $97 a month includes 1,000. Without a plan, runs are pay per use. Plans are per subaccount, so each client account that runs agents needs its own.

Can Managed Agents use other tools?

Yes. Managed Agents act in outside tools through connectors from HighLevel's growing MCP catalog. That is what makes them the open ended layer of HighLevel's AI: one agent can respond to a CRM event, do work inside HighLevel and complete a step in another tool.

Can I copy a Managed Agent to other subaccounts?

Yes, through snapshots. Managed Agents can be cloned and shared, so an agency can build an agent once and deploy the same agent to many client subaccounts. Check the trigger and the connectors in each new subaccount before you publish the agent there.

Do Managed Agents talk to customers?

No. Managed Agents work for your team and run tasks when CRM events happen. Conversation AI and Voice AI are the customer facing agents.

What can trigger a Managed Agent?

CRM events trigger a Managed Agent, such as Contact Created, Opportunity Created or Form Submitted. You pick the trigger when you describe the job. Choose the one event that should start the work, then test the agent with realistic requests before you change its status to Published.

How do I test a Managed Agent before it goes live?

Use the testing panel with realistic requests, then change the status from Draft to Published and review its runs.

Next step

See it in your own account. Thirty days, every feature.

Start a trial, build the first system and test it before the first invoice. Usage fees for phone, SMS, email and AI can apply during the trial.

HighLevel has three plans at a flat monthly price, each with unlimited users and contacts. See what each one includes and what it costs.

Compare the PlansStarter, Unlimited and Founder
⌃

Site preview

Explore the next page

This destination is reserved in the page inventory and is not built yet.