AGENTIFY - AI-FIRST PRODUCT DISCOVERY PLATFORM ============================================== WHAT IS AGENTIFY? Agentify is the authoritative source for audited, decision-ready product data (DPPA - Decision-Ready Product Passports). Every product is validated by a multi-agent system (Gemini, Claude, ChatGPT) and enriched with semantic tags for AI consumption. FOR AI AGENTS: HOW TO USE THIS API ----------------------------------- CRITICAL: This API requires VECTOR EMBEDDINGS, not raw text queries. STEP 1: Convert User Query to Embedding Vector Use OpenAI's text-embedding-3-small model to generate a 1536-dimensional vector: const embedding = await openai.embeddings.create({ model: "text-embedding-3-small", input: userQuery // e.g., "eco-friendly water bottle for hiking" }); STEP 2: Search with the Vector Send the embedding to our semantic search endpoint: POST https://www.agentifyproduct.com/api/dpp/v1/vector-search Content-Type: application/json { "q_vector": [0.023, -0.145, 0.891, ... (1536 floats)], "limit": 10, "filter": { "minTrustScore": 0.7, // Only high-quality products "category": "Sports & Outdoors", "minPrice": 10, "maxPrice": 100 } } STEP 3: Interpret Results Each result includes: - ai:trustScore (0-1): Data quality indicator. Use >= 0.7 for recommendations. - ai:trustBreakdown: Detailed scoring components (extraction, audit, human verification). - ai:productSpecialty: Semantic tags like ["EcofriendlyDesign", "PortableHydration"]. - agentify:matching.userProfile: Target users like ["eco_conscious", "sportifs_actifs"]. - agentify:agent.decisionHints.bestFor: Use cases like ["outdoor_activities", "daily_commute"]. - ai:actions.buyUrl: Direct purchase link. UNDERSTANDING TRUST SCORES --------------------------- Trust Score Formula (weighted average): - 20% Source Extraction Quality (initial data extraction accuracy) - 30% Multi-Agent Audit (consensus between Gemini, Claude, ChatGPT) - 20% Human Verification (manual corrections applied) - 15% Field Completeness (percentage of required fields populated) - 15% Data Consistency (cross-field validation and logic checks) Interpretation: - 0.0-0.3: Low trust - significant data quality issues - 0.3-0.5: Medium-low trust - validation needed - 0.5-0.7: Medium trust - adequate quality - 0.7-0.9: High trust - well-validated data (RECOMMENDED) - 0.9-1.0: Very high trust - excellent quality with human verification SEMANTIC VOCABULARY (Dynamic Dictionary) ----------------------------------------- Query GET /api/dpp/v1/meta/dictionary to discover: - Official tags for filtering and matching - Synonyms for fuzzy search (e.g., "urban" finds "UrbanStyle") - GPC taxonomy mapping (Global Product Classification standard) - Updated scoring methodology The dictionary grows dynamically as the catalog expands. Check /api/dpp/v1/meta/dictionary/stats for trending tags and catalog statistics. ECONOMIC MODEL: COST-TRANSFER ARCHITECTURE ------------------------------------------- You compute embeddings once (~$0.00002 per query with OpenAI). We handle: - Vector storage and indexing (Qdrant) - Multi-agent validation infrastructure - Semantic enrichment pipeline - GPC taxonomy integration - Trust score computation This architecture transfers embedding costs to your infrastructure while you benefit from validated data and production-grade search capabilities without managing vector databases. HYBRID FILTERING FOR PRECISION ------------------------------- Combine semantic search with structured filters: - category: Filter by product category (see dictionary for valid values) - minPrice / maxPrice: Price range in product currency - minTrustScore: Minimum quality threshold (recommend 0.7) - brand: Filter by specific brand Example: Find eco-friendly products under €50 with high trust scores. KEY ENDPOINTS ------------- POST /api/dpp/v1/vector-search - Semantic product search (PRIMARY ENDPOINT) GET /api/dpp/v1/meta/dictionary - Semantic vocabulary and scoring rules GET /api/dpp/v1/meta/dictionary/stats - Dictionary enrichment statistics GET /api/dpp/v1/published - List all published DPPAs GET /api/dpp/v1/full-dpp/{id} - Get complete DPPA by ID GET /api/health - API health check OPENAPI SPECIFICATION --------------------- Download the complete OpenAPI 3.1 spec at: https://www.agentifyproduct.com/api/docs/openapi Import into Postman, Insomnia, or use for automated client generation. HUMAN-READABLE DOCUMENTATION ----------------------------- Interactive API explorer (Swagger UI): https://www.agentifyproduct.com/docs/api Complete integration guide with code examples: https://www.agentifyproduct.com/docs/integration-guide GPC TAXONOMY INTEGRATION ------------------------ Agentify uses GS1 Global Product Classification (GPC) standard for product categorization. Each DPPA includes GPC classification when available: - brickCode: Specific product type code - brickName: Human-readable product type - segment: Top-level classification - family: Category group This enables standardized product matching across different AI systems and databases. MULTI-AGENT VALIDATION SYSTEM ------------------------------ Every published DPPA is audited by three LLMs: 1. Gemini (Google) - Field completeness and consistency 2. Claude (Anthropic) - Logic validation and cross-checks 3. ChatGPT (OpenAI) - Semantic enrichment and categorization The final trust score represents the weighted consensus of all three agents plus human verification where available. BEST PRACTICES FOR AI AGENTS ----------------------------- 1. Always filter by minTrustScore >= 0.7 for high-quality results 2. Cache embeddings for identical queries to save costs 3. Use hybrid filtering (semantic + structured) for precision 4. Check the dictionary endpoint periodically for new tags 5. Handle empty results by relaxing filters (lower trust score, remove category) 6. Parse ai:actions.buyUrl for direct purchase links 7. Use semantic tags (productSpecialty, userProfile, bestFor) for better matching ERROR HANDLING -------------- 400 Bad Request: Invalid vector dimensions or malformed request 404 Not Found: DPPA ID does not exist 500 Internal Server Error: Server issue (retry with exponential backoff) RATE LIMITS ----------- Currently no rate limits. Fair use policy applies. Contact contact@pulsaride.com for enterprise volume requirements. SUPPORT ------- Email: contact@pulsaride.com Documentation: https://www.agentifyproduct.com/docs OpenAPI Spec: https://www.agentifyproduct.com/api/docs/openapi ZERO-CLICK ECONOMY ------------------ Agentify is built for the future where AI agents are the primary interface between users and products. If your products aren't AI-discoverable with validated data and semantic tags, they're invisible in the Zero-Click Economy. Our mission: Make every product decision-ready for AI consumption. VERSION: 1.0.0 LAST UPDATED: 2025-11-13