# Document processing

> Dedicated Flow canvas nodes that apply UiPath document intelligence: extraction, summarization, and more as discrete steps in a process.

Document processing nodes bring UiPath document intelligence directly onto the Flow canvas as discrete, business-oriented steps. Each node represents one operation — extract structured fields, synthesize a long document — without requiring you to wire up a separate agent or understand the underlying model technology.

The nodes appear in the node palette under **Document processing**, grouped by the operation they perform.

## Available nodes

| Node | What it does |
|---|---|
| [Extract](node-extract.md) | Pulls typed fields from a document using a deployed IXP extraction model. Use when you need structured data — invoice totals, form values, named fields — returned in a typed shape for downstream nodes to act on. |
| [Summarize](node-summarize.md) | Synthesizes across a document (up to 1,000 pages) and returns a grounded answer with citations. Use when the next step needs a reasoned answer or summary, not a fixed schema of fields. |

## How document processing nodes differ from agent tools

Summarize also exists as a built-in agent tool that attaches to an [Autonomous Agent](node-agent.md). The distinction is where the decision is made:

- **Document processing node** — the step is deterministic. The flow always calls Extract or Summarize at that point in the process, with the inputs you configure.
- **Agent tool** — the agent decides when to call the tool while reasoning. Use the [Summarize agent tool](summarize.md) when an Autonomous Agent needs to synthesize documents as part of broader open-ended reasoning.

## Related pages

- [Extract](node-extract.md)
- [Summarize](node-summarize.md)
- [Summarize agent tool](summarize.md)
- [Agent](node-agent.md)