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

# Start an async AI workflow

> Initiates an async AI workflow via Temporal (non-streaming fallback). Pre-warms Temporal and MCP connections in parallel. Checks FAQ cache. Validates idempotent client requestIds. Returns 202 immediately with requestId and channelId for Supabase Realtime subscription. Guests are blocked — members, admins, and owners only.




## OpenAPI

````yaml /openapi/ai-brain.yaml post /brain/start
openapi: 3.1.0
info:
  title: Connie AI & Brain API
  version: 1.0.0
  description: >
    API for the AI & Brain module — the orchestration layer powering all AI
    features: the main conversational AI (Brain), AI chat conversations, browser
    automation tasks, agent identity management, and AI usage tracking.

    The Brain is the core AI engine. It accepts messages, orchestrates tool
    calls via Temporal workflows, and streams responses back. It powers the AI
    copilot across the entire product.
servers:
  - url: https://connie.ai/api
    description: Production
security:
  - cookieAuth: []
tags:
  - name: Brain
    description: Core AI engine — conversation, streaming, and workflow orchestration
  - name: Brain Plans
    description: Agent planning context (goal, steps, status)
  - name: AI Chat
    description: Persistent conversation and message management
  - name: AI Chat Attachments
    description: File uploads for AI context
  - name: Browser Tasks
    description: AI-driven browser automation tasks
  - name: Artifacts
    description: KB, table, and workflow artifacts attached to conversations
  - name: Agent Identities
    description: Named AI agent configuration and tool assignments
  - name: AI Cache
    description: FAQ response caching
  - name: AI Usage
    description: Token usage statistics
paths:
  /brain/start:
    post:
      tags:
        - Brain
      summary: Start an async AI workflow
      description: >
        Initiates an async AI workflow via Temporal (non-streaming fallback).
        Pre-warms Temporal and MCP connections in parallel. Checks FAQ cache.
        Validates idempotent client requestIds. Returns 202 immediately with
        requestId and channelId for Supabase Realtime subscription. Guests are
        blocked — members, admins, and owners only.
      operationId: startBrainWorkflow
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/BrainRequest'
      responses:
        '202':
          description: Workflow started
          content:
            application/json:
              schema:
                type: object
                properties:
                  success:
                    type: boolean
                  requestId:
                    type: string
                    format: uuid
                  channelId:
                    type: string
                  status:
                    type: string
                    example: running
                  workflowId:
                    type: string
        '402':
          description: Insufficient credits
        '429':
          description: Rate limited, queue full, or tool-call cap reached (200 max)
components:
  schemas:
    BrainRequest:
      type: object
      required:
        - message
      properties:
        message:
          type: string
        workspaceId:
          type: string
          format: uuid
        intent:
          type: string
          description: Intent classification for routing
        conversationHistory:
          type: array
          maxItems: 100
          items:
            type: object
            properties:
              role:
                type: string
                enum:
                  - user
                  - assistant
                  - tool
              content:
                type: string
        mentions:
          type: array
          items:
            type: object
            properties:
              type:
                type: string
                enum:
                  - mcp
                  - action
                  - workflow
                  - call
              id:
                type: string
              displayName:
                type: string
              description:
                type: string
        knowledgeBaseIds:
          type: array
          items:
            type: string
        knowledgeArticleIds:
          type: array
          items:
            type: string
        tableIds:
          type: array
          items:
            type: string
        fileAttachmentIds:
          type: array
          items:
            type: string
        aiSettings:
          type: object
          properties:
            enabledTools:
              type: array
              items:
                type: string
        pageContext:
          type: object
          properties:
            type:
              type: string
            entityId:
              type: string
            entityName:
              type: string
            url:
              type: string
        reasoning:
          type: string
          enum:
            - 'off'
            - short
            - medium
            - long
        verbosity:
          type: string
          enum:
            - brief
            - medium
            - verbose
        streaming:
          type: boolean
          default: true
        skipCache:
          type: boolean
          default: false
  securitySchemes:
    cookieAuth:
      type: apiKey
      in: cookie
      name: session

````

This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.