Hub Docs User documentation
Back to home
Hub Docs

Agents

What it is

An Agent is an AI automation worker inside your Space. Agents respond to triggers (manual commands, schedules, webhooks, activity events) and take action: creating tasks, updating records, sending notifications, or chaining into other agents.

When to use it

Use agents to automate repetitive work. For example:

  • Run a daily summary of open tasks and email it to the team
  • When a company is created, automatically set up its onboarding tasks
  • When a task is closed, archive related records
  • Accept data from external systems via webhook and create Hub records
  • Generate reports or summaries on demand

Where to find it

Agents are configured in the Tenet admin space. To run an agent manually, use the Ask Anything interface in your Space — type your instruction and the relevant agent handles it. You can also trigger agents via schedules, webhooks, or activity events without opening Chat.

How it works

Using an Agent:

  1. Open Ask Anything in your Space.
  2. Type a prompt or instruction.
  3. The agent processes it and returns a result.
  4. The agent might create tasks, update records, or ask for clarification.

Agent Triggers (Skills):

Agents are triggered by Skills. A single Agent can have multiple Skills:

  • Manual — you invoke it through chat or a button
  • Scheduled — runs on a cron schedule (daily, weekly, monthly)
  • Webhook — triggered by an external system via HTTP
  • Event — triggered by activity in your Space (company created, employee added, etc.)

Agent Actions:

Once triggered, an Agent can:

  • Read records and data
  • Create tasks, comments, records
  • Update existing records
  • Send notifications
  • Trigger other agents
  • Call external APIs via connectors

Agent Results:

When an agent responds with content — a report, a table, a document — it appears inline in the chat as an interactive artifact:

  • Spreadsheet — data displays as an editable grid with column headers, auto-sized columns, and a toolbar. You can sort and interact with it directly.
  • Formatted document — the agent's response renders as rich text with headings, lists, and links.
  • Code — syntax-highlighted code that you can read or copy.

The type of artifact depends on what the agent returned. You don't need to do anything special — Hub renders the result in the right format automatically.

Transparency:

When an Agent runs, it shows you what it did: which records it created, which tasks it assigned, which comments it left. You can always view the execution history.

Related