> ## Documentation Index
> Fetch the complete documentation index at: https://help.iteam.works/llms.txt
> Use this file to discover all available pages before exploring further.

# Agents

> Create, configure, and manage your AI employees. Each agent has a role, skills, tool and data access — and you control it all.

An **agent** is an AI employee. You give it a role, it gains skills and access to the right tools, and it starts executing work when you chat with it or when a routine fires.

<Frame caption="📸 Screenshot to add: Agents list (Intelligence → Agents).">
  <img src="https://mintcdn.com/iteam/E1mg8rFlXblsaNT0/images/placeholder.svg?fit=max&auto=format&n=E1mg8rFlXblsaNT0&q=85&s=36c435f4df28c31a7e114586a931bbd2" alt="Agents list" width="800" height="420" data-path="images/placeholder.svg" />
</Frame>

## Create an agent

<Steps>
  <Step title="Open Intelligence → Agents and click Create agent" />

  <Step title="Describe the role in plain language">
    > "Financial analyst that tracks daily revenue, explains variations, and produces weekly reports."

    iTeam suggests a name, description, skills, and tone of voice.
  </Step>

  <Step title="Adjust whatever you want and save">
    You can refine before saving — or leave it as is and tweak later.
  </Step>
</Steps>

## What you configure in an agent

<AccordionGroup>
  <Accordion title="Identity — name, role, and icon" icon="id-card">
    How the agent appears to your team. Pick an icon that makes it quick to recognize.
  </Accordion>

  <Accordion title="Instructions — the 'personality' and rules" icon="scroll">
    Here you define how it should behave: tone, priorities, what it can and cannot do.
  </Accordion>

  <Accordion title="Skills" icon="puzzle">
    Extra capabilities that expand what the agent can do. [See Skills →](/en/resources/skills)
  </Accordion>

  <Accordion title="Tools and Data Sources" icon="database">
    Which systems and data it can access. [See Data Sources →](/en/resources/data-sources)
  </Accordion>

  <Accordion title="AI model" icon="cpu">
    By default iTeam picks the best model for each request. You can pin one if you want.
  </Accordion>

  <Accordion title="Access (who can use it)" icon="lock">
    Define who can see and use this agent. [See Security & Access →](/en/administration/security-and-access)
  </Accordion>

  <Accordion title="Agent widgets" icon="layout-dashboard">
    The interactive screens it has built. [See Widgets →](/en/chat/widgets)
  </Accordion>
</AccordionGroup>

## Best practices

<CardGroup cols={2}>
  <Card title="One clear role per agent" icon="bullseye">
    Focused agents answer better. Prefer "Sales Analyst" over "does everything".
  </Card>

  <Card title="Give context in the instructions" icon="book">
    The more it knows about your company, the sharper the answers.
  </Card>

  <Card title="Less is more on tools" icon="filter">
    Grant only the access needed — it's safer and faster.
  </Card>

  <Card title="Evolve with use" icon="trending-up">
    Tune the instructions as you learn what works.
  </Card>
</CardGroup>

<Info>
  iTeam **learns from use**: the agent remembers what's already been done and gets more accurate over time, without you repeating context.
</Info>

## Next steps

<CardGroup cols={2}>
  <Card title="Agent Teams" icon="users" href="/en/intelligence/agent-teams">
    Group agents that work together.
  </Card>

  <Card title="Chat" icon="messages-square" href="/en/chat/chat-with-agents">
    Put your agent to work.
  </Card>
</CardGroup>