# AI agents in BRIX AI

> [HTML Version](cortex-ai-agents.html)

AI agents in **BRIX AI** are intelligent assistants powered by large language models. Employees can use an agent as an AI assistant to chat, request information, and delegate tasks in BRIX.

You can create dedicated AI agents for different business goals and departments, such as tech support, sales, accounting, and more.

````
начало внимание

````
Creating AI agents is available only in **BRIX AI Standard **and** BRIX AI Enterprise**. Learn more about the features available in your [edition](cortex-editions.md).

````
конец внимание

````
To configure AI agent behavior, use:

- Language model prompts: Text instructions that define how the AI agent understands tasks and formats its responses.

- Tools: Functions that an AI agent uses to retrieve information, process data, and perform actions in the system. Available options include:

	- Built-in tools for working with date and time, external website content, files, images, and BRIX data.

	- **BRIX tools** that are custom functions for interacting with specific company apps, processes, and tasks.

	- Smart search, other agents, or AI actions connected as tools. To get accurate results, you can route tasks between **BRIX AI** components and invoke components configured for specialized tasks.

	- Custom tools connected via MCP servers that are custom functions you can set up, such as fetching data from external systems.

## Task routing for AI agents

In **BRIX AI**, you can set up flexible task distribution between AI agents. You can create main AI agents that interact with users and child AI agents to which main agents pass specific tasks. This helps route requests based on specialization to achieve more accurate results.

For example, a **Sales Assistant** AI agent can delegate reviewing a contract to one child agent and drafting a customized commercial proposal to another. Each child agent is assigned relevant instructions and tools.

You can add a child AI agent to a main agent:

- As a tool. The child AI agent receives only the input data passed by the main AI agent, processes it, and returns the result.

- As a managed AI agent. The child AI agent receives full context of the user interaction from the main AI agent to complete the task with all relevant information.

To learn more about AI agents, see:

- [Create and configure an AI agent](cortex-configure-agent.md).

- [Debug an AI agent](cortex-agent-calibration.md).

- [Configure employee access to AI agents in BRIX](cortex-agent-access.md).

- [Work with AI agents in chat](cortex-work-with-agents.md).