Vectorize Extraction API

openapi: 3.0.1
info:
  title: Vectorize AI Platform Connectors Extraction API
  description: 'The Vectorize API lets developers programmatically build and operate RAG (retrieval-augmented generation) pipelines: create / start / stop / delete pipelines, run retrieval against a pipeline''s vector index, manage source, destination, and AI-platform connectors, upload files, extract documents with Iris, and run deep research. All paths are organization-scoped under /org/{organizationId}. Authentication uses a Bearer access token.'
  termsOfService: https://vectorize.io/terms
  contact:
    name: Vectorize Support
    url: https://vectorize.io
  version: 0.1.2
servers:
- url: https://api.vectorize.io/v1
  description: Vectorize API
security:
- bearerAuth: []
tags:
- name: Extraction
  description: Vectorize Iris document extraction.
paths:
  /org/{organizationId}/extraction:
    post:
      operationId: startExtraction
      tags:
      - Extraction
      summary: Start content extraction from a file
      description: Starts a Vectorize Iris extraction job against a previously uploaded file, converting it into Markdown chunks with optional metadata extraction.
      parameters:
      - $ref: '#/components/parameters/OrganizationId'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/StartExtractionRequest'
      responses:
        '200':
          description: Extraction started.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/StartExtractionResponse'
        '400':
          $ref: '#/components/responses/Error'
  /org/{organizationId}/extraction/{extractionId}:
    get:
      operationId: getExtractionResult
      tags:
      - Extraction
      summary: Get extraction result
      parameters:
      - $ref: '#/components/parameters/OrganizationId'
      - name: extractionId
        in: path
        required: true
        schema:
          type: string
      responses:
        '200':
          description: Extraction result. Poll until ready is true.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ExtractionResultResponse'
        '404':
          $ref: '#/components/responses/Error'
components:
  schemas:
    ErrorResponse:
      type: object
      properties:
        error:
          type: string
        details:
          type: string
        failedUpdates:
          type: array
          items:
            type: string
        successfulUpdates:
          type: array
          items:
            type: string
      required:
      - error
    ExtractionResultResponse:
      type: object
      properties:
        ready:
          type: boolean
        data:
          $ref: '#/components/schemas/ExtractionResult'
      required:
      - ready
    StartExtractionResponse:
      type: object
      properties:
        message:
          type: string
        extractionId:
          type: string
      required:
      - message
      - extractionId
    StartExtractionRequest:
      type: object
      properties:
        fileId:
          type: string
        type:
          type: string
          enum:
          - iris
          default: iris
        chunkingStrategy:
          type: string
          enum:
          - markdown
          default: markdown
        chunkSize:
          type: number
          default: 256
        metadata:
          type: object
          description: Metadata extraction strategy.
          additionalProperties: true
      required:
      - fileId
    ExtractionResult:
      type: object
      properties:
        success:
          type: boolean
        chunks:
          type: array
          items:
            type: string
        text:
          type: string
        metadata:
          type: string
        metadataSchema:
          type: string
        chunksMetadata:
          type: array
          items:
            type: string
        chunksSchema:
          type: array
          items:
            type: string
        error:
          type: string
      required:
      - success
  parameters:
    OrganizationId:
      name: organizationId
      in: path
      required: true
      description: Vectorize organization identifier.
      schema:
        type: string
  responses:
    Error:
      description: Error response.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/ErrorResponse'
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      bearerFormat: JWT
      description: Vectorize API access token passed as a Bearer token.