Ai Search Competitors

AI-powered competitor search using Bedrock Claude + Tavily.

Cache-First Architecture Enhancement

This endpoint searches for competitors using AI when cache is empty or stale. Results are automatically cached for fast subsequent queries.

How It Works:

1. Uses Tavily to search web for "{vertical} companies 2024 2025"
2. Claude Sonnet 4 extracts structured competitor list
3. Results are cached in brand_library table
4. Next query for same vertical returns cached results instantly

Example Request:

POST /api/v1/brand-library/ai-search-competitors?vertical=Agriculture%20Tech&limit=15

Example Response:

{
  "vertical": "Agriculture Tech",
  "count": 15,
  "total_brands": 15,
  "suggestions": [
    {
      "brand": "John Deere",
      "tier": "Leader",
      "rank": 1,
      "confidence": "High",
      "signal": "Leading precision agriculture with autonomous tractors in 2024",
      "is_parent": false,
      "products": []
    },
    ...
  ]
}

Args:

• vertical: Industry vertical to search (e.g., "Agriculture Tech")
• limit: Max competitors to find (1-50, default 15)
• cache_results: Whether to save results to database (default true)

Returns:

• List of AI-discovered competitors with metadata

Status Codes:

• 200: Success (may take 3-10 seconds for AI search, returns empty list if Tavily API fails)
• 500: AI search failed (check logs)

Performance:

• First query: 3-10 seconds (AI search + web scraping)
• Cached queries: <100ms (instant from database)

Fallback Behavior:

• If Tavily API is unavailable (quota exceeded, rate limit, network error), returns empty results
• Clients should fall back to cached data from /suggest-competitors endpoint
• Graceful degradation: no errors exposed to end users, just empty competitor list
• Tavily API errors logged as WARNING for monitoring