AI Chat API (RAG)

Query your contracts, knowledge base, and documents using AI-powered retrieval-augmented generation.

Endpoints

Method	Path	Description
`POST`	`/chat/send`	Send a message
`GET`	`/chat/history`	Get conversation history
`POST`	`/chat/sync-contracts`	Sync contracts to vector store
`GET`	`/chat/stats`	Get RAG system stats

Send Message

POST /api/v1/chat/send

{
  "message": "What are the termination clauses in our Acme NDA?",
  "conversation_id": "optional-conv-uuid",
  "stream": true
}

Field	Type	Required	Description
`message`	string	✅	Your question
`conversation_id`	string	—	Continue existing conversation
`stream`	boolean	—	Enable SSE streaming (default: true)

Streaming Response (Server-Sent Events):

data: {"type": "chunk", "content": "The Acme NDA contains"}
data: {"type": "chunk", "content": " a 30-day termination clause"}
data: {"type": "chunk", "content": " in Section 7.2."}
data: {"type": "citations", "sources": [{"title": "NDA - Acme Corp", "page": 4, "relevance": 0.89}]}
data: {"type": "done", "conversation_id": "conv-uuid"}

The AI searches across 3 collections: your contracts, knowledge base documents, and uploaded user documents — all scoped to your organization.

How RAG Works

User Question
    ↓
Embed with text-embedding-3-large (3072 dims)
    ↓
Search Qdrant (3 collections × org_id filter)
    ↓
Filter by relevance score ≥ 0.35
    ↓
Deduplicate overlapping chunks
    ↓
Build context (max 28,000 chars)
    ↓
GPT-5.1 generates answer with citations
    ↓
Stream response via SSE

Get Conversation History

GET /api/v1/chat/history?conversation_id=conv-uuid

Response:

{
  "conversation_id": "conv-uuid",
  "messages": [
    {
      "role": "user",
      "content": "What are the termination clauses in our Acme NDA?",
      "timestamp": "2026-03-08T12:00:00Z"
    },
    {
      "role": "assistant",
      "content": "The Acme NDA contains a 30-day termination clause...",
      "citations": [
        {
          "title": "NDA - Acme Corp",
          "page": 4,
          "section": "Section 7.2",
          "relevance": 0.89
        }
      ],
      "timestamp": "2026-03-08T12:00:03Z"
    }
  ]
}

Sync Contracts

Index all organization contracts into the vector store for AI search.

POST /api/v1/chat/sync-contracts

⚠️

This endpoint re-indexes all contracts. Use after bulk uploads or when first setting up.

RAG Stats

GET /api/v1/chat/stats

{
  "collections": {
    "contracts_kb": {"vectors": 1250, "status": "green"},
    "knowledge_base": {"vectors": 340, "status": "green"},
    "user_documents": {"vectors": 89, "status": "green"}
  },
  "models": {
    "primary": "gpt-5.1",
    "mini": "gpt-5-mini",
    "premium": "gpt-5.4",
    "fallback": "claude-sonnet-4-20250514",
    "embedding": "text-embedding-3-large"
  },
  "config": {
    "embedding_dim": 3072,
    "min_relevance_score": 0.35,
    "max_context_chars": 28000
  }
}

Signatures Compliance