deepset-ai--haystack/docs-website/docs/concepts/agents.mdx

---
title: "Agents"
id: agents
slug: "/agents"
description: "This page explains how to create an AI agent in Haystack capable of retrieving information, generating responses, and taking actions using various Haystack components."
---

# Agents

This page explains how to create an AI agent in Haystack capable of retrieving information, generating responses, and taking actions using various Haystack components.

## What’s an AI Agent?

An AI agent is a system that can:

- Understand user input (text, image, audio, and other queries),
- Retrieve relevant information (documents or structured data),
- Generate intelligent responses (using LLMs like OpenAI or Hugging Face models),
- Perform actions (calling APIs, fetching live data, executing functions).

AI agents are autonomous systems that use large language models (LLMs) to make decisions and solve complex tasks.
They interact with their environment using tools, memory, and reasoning.
An AI agent is more than a chatbot — it actively plans, chooses the right tools, and executes tasks to achieve a goal.
Unlike traditional software, it adapts to new information and refines its process as needed.

1. **LLM as the Brain**: The agent’s core is an LLM, which understands context, processes natural language and serves as the central intelligence system.
2. **Tools for Interaction**: Agents connect to external tools, APIs, and databases to gather information and take action.
3. **Memory for Context**: Short-term memory helps track conversations, while long-term memory stores knowledge for future interactions.
4. **Reasoning and Planning**: Agents break down complex problems, come up with step-by-step action plans, and adapt based on new data and feedback.

An AI agent starts with a prompt that defines its role and objectives.
It decides when to use tools, gathers data, and refines its approach through loops of reasoning and action.
For example, a customer service agent answers queries using a database — if it lacks an answer, it fetches real-time data, summarizes it, and responds.
A coding assistant understands project requirements, suggests solutions, and writes code.

## Key Components

### Agent Component

Haystack has a built-in [Agent](../pipeline-components/agents-1/agent.mdx) component that manages the full tool-calling loop — it calls the LLM, invokes tools, updates state, and continues until a stopping condition is met.
Key capabilities include:

- **State management**: Share typed data between tools, accumulate results across iterations, and surface them in the result dict using `state_schema`. See [State](../pipeline-components/agents-1/state.mdx).
- **Streaming**: Stream token-by-token output with a `streaming_callback`.
- **Human-in-the-loop**: Intercept tool calls for human review before execution. See [Human in the Loop](../pipeline-components/agents-1/human-in-the-loop.mdx).
- **Multi-agent systems**: Wrap an `Agent` as a `ComponentTool` to build coordinator/specialist architectures. See [Multi-Agent Systems](./agents/multi-agent-systems.mdx).
- **MCP server exposure**: Expose your agent as an MCP server using [Hayhooks](../development/hayhooks.mdx), making it callable from any MCP-compatible client such as Claude Desktop or Cursor.
- **Multimodal inputs**: Pass images alongside text using `ImageContent` in `ChatMessage` content parts, or return `ImageContent` from tools for dynamic image analysis. Requires a vision-capable model such as `gpt-5` or `gemini-2.5-flash`. See [Multimodal Inputs](../pipeline-components/agents-1/agent.mdx#multimodal-inputs).

Check out the [Agent](../pipeline-components/agents-1/agent.mdx) documentation, or the [example](#tool-calling-agent) below to get started.

### State

[`State`](../pipeline-components/agents-1/state.mdx) is Haystack's built-in mechanism for sharing data between tools and accumulating results across multiple tool calls.
You define a `state_schema` on the `Agent`, and any keys declared there are returned alongside `messages` and `last_message` in the agent's result dict.

### Tools

Haystack provides several ways to create and manage tools:

- [`Tool`](../tools/tool.mdx) class / [`@tool`](../tools/tool.mdx#tool-decorator) decorator – Define a tool from a Python function. The `@tool` decorator automatically uses the function's name and docstring; the `Tool` class gives full control over the name, description, and schema.
- [`ComponentTool`](../tools/componenttool.mdx) – Wraps any Haystack component as a callable tool.
- [`PipelineTool`](../tools/pipelinetool.mdx) – Wraps a full Haystack pipeline as a callable tool.
- [`MCPTool`](../tools/mcptool.mdx) / [`MCPToolset`](../tools/mcptoolset.mdx) – Connects to Model Context Protocol (MCP) servers to load external tools.
- [`Toolset`](../tools/toolset.mdx) – Groups multiple tools into a single unit to pass to an Agent or Generator.
- [`SearchableToolset`](../tools/searchabletoolset.mdx) – Enables keyword-based tool discovery for large catalogs, so the LLM only sees relevant tools at each step.

## Example

### Tool-Calling Agent

Create a tool-calling agent with the `Agent` component. This example requires `OPENAI_API_KEY` and `SERPERDEV_API_KEY` to be set as environment variables:

```shell
export OPENAI_API_KEY=<your-openai-key>
export SERPERDEV_API_KEY=<your-serperdev-key>
```

The examples on this page use SerperDev web search component that have moved to the `serperdev-haystack` package. Install it to run the examples:

```shell
pip install serperdev-haystack
```

```python
from haystack.components.agents import Agent
from haystack.components.generators.chat import OpenAIChatGenerator
from haystack.components.generators.utils import print_streaming_chunk
from haystack_integrations.components.websearch.serperdev import SerperDevWebSearch
from haystack.dataclasses import ChatMessage
from haystack.tools import ComponentTool

# Wrap the web search component as a tool
web_tool = ComponentTool(
    component=SerperDevWebSearch(top_k=3),
    name="web_search",
    description="Search the web for current information like weather, news, or facts.",
)

tool_calling_agent = Agent(
    chat_generator=OpenAIChatGenerator(model="gpt-5.4-nano"),
    system_prompt=(
        "You're a helpful agent. When asked about current information like weather, news, or facts, "
        "use the web_search tool to find the information and then summarize the findings."
    ),
    tools=[web_tool],
    streaming_callback=print_streaming_chunk,
)

result = tool_calling_agent.run(
    messages=[ChatMessage.from_user("How is the weather in Berlin?")],
)
print(result["last_message"].text)
```

Resulting in:

```python
>>> The current weather in Berlin is approximately 60°F. The forecast for today includes clouds in the morning with some sunshine later. The high temperature is expected to be around 65°F, and the low tonight will drop to 40°F.

- **Morning**: 49°F
- **Afternoon**: 57°F
- **Evening**: 47°F
- **Overnight**: 39°F

For more details, you can check the full forecasts on [AccuWeather](https://www.accuweather.com/en/de/berlin/10178/current-weather/178087) or [Weather.com](https://weather.com/weather/today/l/5ca23443513a0fdc1d37ae2ffaf5586162c6fe592a66acc9320a0d0536be1bb9).
```