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

# AI Nodes

> Generate text, analyze content, extract data, and make decisions with AI

Innflow includes dedicated AI nodes for common automation patterns and flexible model nodes for custom prompts. AI nodes can use data from earlier workflow steps through [template variables](/essentials/variables).

## AI Agent

The AI Agent node runs an agentic loop with tool calling. Use it when the model needs to reason across multiple steps, call connected tools, or decide which operation to run next.

**Configuration:**

| Field             | Description                                                 |
| ----------------- | ----------------------------------------------------------- |
| **Variable name** | Name used to reference the output in downstream nodes       |
| **Instructions**  | System-level guidance for the agent                         |
| **Input**         | User message or task payload, often with template variables |
| **Tools**         | Tool nodes connected to the agent                           |
| **Model**         | The model used for reasoning and tool calls                 |

**Accessing output:** `{{variableName.output.text}}`

***

## AI Model

The AI Model node is the general purpose LLM node. Use it for summarization, classification, rewriting, extraction, drafting, and other prompt-based work.

**Configuration:**

| Field             | Description                                               |
| ----------------- | --------------------------------------------------------- |
| **Variable name** | Name used to reference the output in downstream nodes     |
| **Model**         | Choose the model for the task                             |
| **Messages**      | System, user, and assistant messages to send to the model |
| **Temperature**   | Controls response variability when available              |

**Example prompt:**

```txt theme={null}
Summarize this support request and identify the next action:
{{gmail.output.body}}
```

***

## Provider Nodes

Provider-specific nodes are available when you want to route work to a specific model family or provider configuration.

| Node          | Use for                                           |
| ------------- | ------------------------------------------------- |
| **OpenAI**    | General text generation and structured reasoning  |
| **Anthropic** | Long-form analysis and careful drafting           |
| **Gemini**    | Fast multimodal-friendly generation and reasoning |
| **DeepSeek**  | Cost-conscious reasoning and coding tasks         |
| **Kimi**      | Moonshot/Kimi model access                        |
| **Grok**      | xAI model access                                  |
| **Qwen**      | Alibaba Cloud/Qwen model access                   |

***

## Label Fields

The Label Fields node extracts structured data from unstructured text. Define the fields you want to extract and their types, then use the extracted values in later nodes.

**Configuration:**

| Field             | Description                               |
| ----------------- | ----------------------------------------- |
| **Variable name** | Name used to reference the extracted data |
| **Input**         | The text to extract data from             |
| **Fields**        | Field names and expected types            |

**Example:** Extract `name`, `email`, and `sentiment` from a customer support email.

**Accessing output:** `{{variableName.output.name}}`, `{{variableName.output.email}}`

***

## Router

The Router node uses AI to choose the best branch for an input. Define named routes with descriptions, then connect each route to the workflow path that should run.

Router configuration is covered in [Logic and Utility](/tools/logic-utility#router).