> ## Documentation Index
> Fetch the complete documentation index at: https://thethirdpenco-feat-tool-lifecycle-events.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Navigating ai-query

> Which API to use first, and when to move deeper

This page is the map for the framework.

`ai-query` is intentionally layered. You should not learn everything at once.

## Start Here

If you are new, follow this order:

1. [Quickstart](/tutorials/quickstart)
2. [Concepts](/core/overview)
3. [Generating Text](/core/generate-text)
4. [Streaming](/core/streaming)
5. [Tools](/core/tools)
6. [Stateful Agents](/core/stateful-agents)
7. [Live Turns](/core/live-turns)

## The Decision Tree

### I just want to call a model

Use `generate_text()`.

```python theme={null}
result = await generate_text(model=model, prompt="Hello")
```

Read: [Generating Text](/core/generate-text)

### I want streamed model output

Use `stream_text()`.

```python theme={null}
result = stream_text(model=model, prompt="Hello")
async for chunk in result.text_stream:
    print(chunk, end="")
```

Read: [Streaming](/core/streaming)

### I want the model to call Python functions

Use tools.

```python theme={null}
result = await generate_text(
    model=model,
    prompt="What's the weather?",
    tools={"get_weather": get_weather},
)
```

Read: [Tools](/core/tools)

### I want memory or sessions

Use `Agent`.

```python theme={null}
agent = Agent("assistant", model=model, storage=SQLiteStorage("agents.sqlite3"))
text = await agent.chat("Remember this")
```

Read: [Stateful Agents](/core/stateful-agents)

### I want metadata about an agent response

Use `Agent.run()`.

```python theme={null}
result = await agent.run("Investigate this")
print(result.steps)
```

Read: [Agent Reference](/reference/agent)

### I am building a CLI or web UI

Use `Agent.turn()`.

```python theme={null}
turn = agent.turn("Do the task")
async for event in turn.events():
    ...
```

Read: [Live Turns](/core/live-turns)

### I want to serve many users

Use `AgentServer`.

```python theme={null}
AgentServer(UserAgent).serve(port=8080)
```

Read: [Agent Server](/core/agent-server)

## Import Map

Common imports:

```python theme={null}
from ai_query import generate_text, stream_text, tool, Field
from ai_query.agents import Agent, AgentServer, SQLiteStorage
from ai_query.providers import openai, anthropic, google
```

Core types and advanced control:

```python theme={null}
from ai_query import StepControl, step_count_is, has_tool_call
from ai_query.agents import AgentTurn, TurnOptions, TurnResult
from ai_query.types import AbortSignal, Message
```

## Product Boundaries

These are the boundaries to keep in your head:

* Providers create models.
* Core functions run model calls.
* Tools give models capabilities.
* Agents add identity, state, storage, and message history.
* Turns expose one live agent execution.
* Servers expose agents over HTTP, SSE, WebSocket, and RPC.

If a page does not fit one of those categories, it probably belongs in the Cookbook or API Reference.