Skip to content
Hot Local Business
Start onboarding

Agent studio

Build custom multi-step AI agents with tools, skills and knowledge.

Why it matters

Create AI helpers for specific jobs.

The HLB difference

Agents can be deployed to channels or via API.

How it works

Build custom AI agents with their own tools and knowledge for specific jobs in your business.

  • Set a managed agent to run on its own when a contact or opportunity is created, or when an opportunity status changes, so lead qualification, sales follow-ups and customer onboarding can start without a manual step.
  • Describe the agent you want in plain language and the guided builder asks clarifying questions, makes targeted edits and shows the setup in a live panel you can adjust, so you can update an agent without rebuilding it.
  • Open a managed agent in agent studio, type the change you want in plain words, then review and fine-tune the result in the live configuration panel.
  • In agent studio, open the trigger picker on a managed agent, choose contact created, opportunity created or opportunity status changed, and set any pipeline, stage or status filters.
  • Open a managed agent's instructions, choose Insert value and resolve any missing values before publishing.

Build and edit AI agents by describing what you want

Tweak your AI helper between jobs in a minute or two, without rebuilding it or learning a settings screen.

In agent studio, you can now create or change a managed agent by simply telling it what you want, like "make the tone friendlier" or "add a trigger for the contact form." It changes only what you asked for, asks a quick question when a request is unclear, and shows each change in a live panel you can edit yourself. It also works with the forms, calendars, pipelines and knowledge bases already in your account.

See also Conversation AI, AI knowledge bases and Forms.

Start AI agents when a new lead or deal comes in

New leads get a prompt first touch and won jobs move into onboarding, even on the days your team is slammed.

Managed agents can now run on their own when a new contact is added, when a new opportunity is created, or when an opportunity changes status. Filter by pipeline, stage and status so the right agent handles welcome messages, lead qualification or follow-up after a job is won or lost.

See also Opportunity pipelines, Contact records and Visual workflow builder.

Reuse business details in your AI instructions

Update a booking link or office email in one place so your agent can use the current details on its next run.

Insert saved custom values and business profile details into managed agent instructions in HLB. When you update a saved value, the agent reads the current information the next time it runs, without needing to be republished. Missing referenced values are flagged before publishing.

See also AI knowledge bases and Conversation AI.

Agent studio lets you build custom AI agents that answer questions, search your business information, collect details and take action. New agents are managed agents, set up with instructions, skills, knowledge sources and tools. Agents built with the older visual flow builder keep working and stay editable without rebuilding their connected automations. For those, saved changes go to a staging version until you promote it to production.

For a local service business, an agent might answer common questions from your approved FAQs and policies, or update contact records within a defined process.

Built-in capabilities let an agent send messages, update CRM data, apply tags, book appointments and start processes, depending on the agent type. Custom skills hold saved instructions for a repeatable task and can be toggled per managed agent. Knowledge sources give the agent approved information to answer from. Supported agents can also reach outside systems through integrations or MCP connectors, and custom API calls use GET, POST, PUT, PATCH or DELETE requests. You test each request and choose which response fields the agent uses.

Ways to run an agent:

  • Chat, for direct conversations.
  • Workflows, using the Invoke Managed Agent action, so later steps can use the agent's response.
  • A schedule, once at a set time, at a repeating interval or with a cron expression.
  • The API, so your website or other software can call the published agent with a private integration token, using ready-made cURL, Python or JavaScript requests.

Enable only the capabilities the agent needs, test with realistic requests, then publish. API calls always use the published version, and the Activity tab shows each run with the skills and tools it used.

Related features

Ready to put this to work in your business?

Start onboarding with the Max Digital Edge team and get set up on Hot Local Business.

Start onboarding