Synthesis

Overview

Synthesis is the core AI workflow in Praxiom AI. It takes selected research sources (interviews, surveys, support tickets) and uses the Claude Agent SDK to cross-reference themes, extract structured insights with quotes, severity, and frequency, then persists them to the database.

---

Synthesize Research

POST /api/research/synthesize

Trigger AI synthesis of research sources into structured insights. This is a blocking endpoint that typically takes 15-60 seconds to complete. For real-time progress updates, use the SSE streaming endpoint instead.

Requires: Active subscription. Rate-limited to 10 requests per minute.

Request Body

Name	Type	Required	Description
workspace_id	UUID	Yes	Target workspace
source_ids	UUID[]	Yes	List of research source IDs to synthesize
synthesis_type	string	No	Type of synthesis (default: "comprehensive")

Prerequisites

• All referenced sources must have processing_status: "completed"
• No other synthesis can be running for the same workspace (returns 409)

Example Request

curl -X POST https://api.praxiomai.xyz/api/research/synthesize \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "workspace_id": "550e8400-e29b-41d4-a716-446655440000",
    "source_ids": [
      "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
      "b2c3d4e5-f6a7-8901-bcde-f12345678901"
    ],
    "synthesis_type": "comprehensive"
  }'

import requests

resp = requests.post(
    "https://api.praxiomai.xyz/api/research/synthesize",
    headers={"Authorization": f"Bearer {token}"},
    json={
        "workspace_id": "550e8400-e29b-41d4-a716-446655440000",
        "source_ids": [
            "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
            "b2c3d4e5-f6a7-8901-bcde-f12345678901",
        ],
        "synthesis_type": "comprehensive",
    },
)

const resp = await fetch("https://api.praxiomai.xyz/api/research/synthesize", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${token}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    workspace_id: "550e8400-e29b-41d4-a716-446655440000",
    source_ids: [
      "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
      "b2c3d4e5-f6a7-8901-bcde-f12345678901",
    ],
    synthesis_type: "comprehensive",
  }),
});

Response (200 OK)

{
  "insights": [
    {
      "id": "c3d4e5f6-a7b8-9012-cdef-123456789012",
      "workspace_id": "550e8400-e29b-41d4-a716-446655440000",
      "insight_type": "pain_point",
      "title": "Users struggle with onboarding flow",
      "description": "Multiple interview subjects mentioned confusion during initial setup...",
      "source_ids": ["a1b2c3d4-e5f6-7890-abcd-ef1234567890"],
      "quotes": [
        {
          "text": "I had no idea what to do after signing up",
          "source_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
          "speaker": "User 3"
        }
      ],
      "frequency": 4,
      "severity": "high",
      "created_at": "2026-03-27T10:35:00Z"
    }
  ],
  "insight_ids": ["c3d4e5f6-a7b8-9012-cdef-123456789012"],
  "summary": "Synthesized 5 insights from 2 sources. Key themes: onboarding friction, pricing confusion, and feature discoverability."
}