RAG Search API

Search your knowledge base using semantic similarity. The RAG (Retrieval Augmented Generation) API finds relevant content that can be used to ground AI responses.

POST /api/sdk/rag/search

Search for relevant content using natural language queries.

Request Body

{
  query: string;                   // Search query

  // Optional
  searchType?: 'hybrid' | 'standard' | 'expanded';
  maxResults?: number;             // Default: 10, max: 50
  threshold?: number;              // Minimum relevance score (0-1)
  useReranking?: boolean;          // Re-rank results for relevance
  contentTypes?: string[];         // Filter by content type
  context?: Record<string, any>;   // Additional context for search
}

Search Types

| Type | Description | Best For | |------|-------------|----------| | standard | Basic semantic search | Simple queries | | hybrid | Combines semantic + keyword | Most use cases | | expanded | Query expansion + hybrid | Complex queries |

Example Request

curl -X POST https://app.fig1.ai/api/sdk/rag/search \
  -H "Content-Type: application/json" \
  -H "X-Fig1-API-Key: fig1_sdk_abc123" \
  -d '{
    "query": "How do I reset my password?",
    "searchType": "hybrid",
    "maxResults": 5,
    "contentTypes": ["faq", "article"]
  }'

Response

{
  "success": true,
  "data": {
    "results": [
      {
        "id": "chunk_abc123",
        "content": "To reset your password, go to the login page and click 'Forgot Password'. Enter your email address and we'll send you a reset link...",
        "score": 0.92,
        "rank": 1,
        "source": "faq",
        "metadata": {
          "fileName": "password-reset.md",
          "contentType": "faq",
          "title": "Password Reset Guide"
        }
      },
      {
        "id": "chunk_def456",
        "content": "If you're having trouble with your account credentials, our password reset feature can help...",
        "score": 0.85,
        "rank": 2,
        "source": "article",
        "metadata": {
          "fileName": "account-troubleshooting.md",
          "contentType": "article",
          "title": "Account Troubleshooting"
        }
      }
    ],
    "totalResults": 2,
    "query": "How do I reset my password?",
    "searchType": "hybrid",
    "latencyMs": 145
  }
}

Using RAG with Chat

The agent automatically uses RAG when configured. You can also provide search results as context:

// 1. Search for relevant content
const searchResponse = await fetch('https://app.fig1.ai/api/sdk/rag/search', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'X-Fig1-API-Key': 'fig1_sdk_abc123'
  },
  body: JSON.stringify({
    query: 'password reset',
    maxResults: 3
  })
});

const { data: searchData } = await searchResponse.json();

// 2. Use results as context in chat
const chatResponse = await fetch('https://app.fig1.ai/api/sdk/agent/chat', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'X-Fig1-API-Key': 'fig1_sdk_abc123'
  },
  body: JSON.stringify({
    message: 'How do I reset my password?',
    context: {
      relevantContent: searchData.results.map(r => r.content)
    }
  })
});

Knowledge Base Management

GET /api/sdk/rag/documents

List documents in your knowledge base.

curl -X GET "https://app.fig1.ai/api/sdk/rag/documents" \
  -H "X-Fig1-API-Key: fig1_sdk_abc123"

Response:

{
  "success": true,
  "data": [
    {
      "_id": "doc_abc123",
      "fileName": "product-guide.pdf",
      "fileType": "pdf",
      "status": "completed",
      "chunks": 45,
      "createdAt": "2024-01-15T10:00:00Z",
      "processedAt": "2024-01-15T10:02:30Z"
    }
  ]
}

Document Status

| Status | Description | |--------|-------------| | queued | Waiting to be processed | | processing | Currently being analyzed and indexed | | completed | Successfully indexed and searchable | | failed | Processing failed (check error message) |

Indexing Content

Documents are automatically chunked and indexed when uploaded. The system:

1. Extracts text from documents (PDF, DOCX, etc.)
2. Splits content into semantic chunks
3. Generates embeddings using Voyage AI
4. Stores in vector database for fast retrieval

Supported File Types

| Type | Extensions | Notes | |------|------------|-------| | PDF | .pdf | Text extraction + OCR | | Word | .docx, .doc | Full formatting preserved | | Markdown | .md | Headers become sections | | Text | .txt | Plain text | | HTML | .html | Tags stripped |

Best Practices

1. Use hybrid search - Best balance of semantic and keyword matching
2. Set appropriate thresholds - Filter out low-relevance results
3. Enable re-ranking - Improves result quality for important queries
4. Structure your content - Clear headings and sections improve chunking
5. Keep content updated - Re-index when source content changes