---
title: "Run an Extraction node (LLM)"
method: POST
path: "/v1/nodes/extract"
tags: ["Nodes"]
---

# Run an Extraction node (LLM)

`POST /v1/nodes/extract`

## Request

- Base URL: `https://api.talonic.com`
- URL: `https://api.talonic.com/v1/nodes/extract`
- Auth: HTTP bearer

LLM-extract a schema's fields. Start a fresh canvas with `document_ids`, or extract into an existing `record_set_id` (e.g. after a transfer node — already filled fields are skipped). Provide exactly one. Optional `field_keys` scopes extraction to a subset of the schema.

## Request body

- object
  - `schema_id` string, uuid, required
  - `document_ids` string[]
  - `record_set_id` string, uuid
  - `field_keys` string[]

## Response `202`

Node job created and queued

- NodeRun: A standalone run of one One Engine phase over a record set.
  - `id` string, uuid
  - `node_type` 'transfer' | 'extraction' | 'resolution' | 'validation' | 'assembly'
  - `status` 'queued' | 'running' | 'completed' | 'partial' | 'error'
  - `record_set_id` string, uuid, nullable
  - `product_record_set_id` string, uuid, nullable
  - `total` integer
  - `completed` integer
  - `errors` integer
  - `error_message` string, nullable
  - `created_at` string, date-time
  - `links` object

## Other responses

- `400`: Invalid request (provide either document_ids or record_set_id)
- `404`: Schema

---

[API](https://skmtc.dev/talonic/apis/talonic-api.md) · [All operations](https://skmtc.dev/talonic/apis/talonic-api/llms.txt) · [OpenAPI document](https://skmtc.dev/talonic/apis/talonic-api/revisions/678fe5f8c9c3?raw)