Recommendation Engine
Personalized AI recommendations with user history
About
Build a personalized recommendation engine powered by vector similarity and user interaction history. Ingest your product or content catalog, track user views/purchases/ratings, and get AI-ranked recommendations with natural-language explanations.
The engine combines semantic similarity (via vector search) with collaborative signals (stored persistently) to surface trending items and personalized picks.
API Endpoints
/ingest-items /recommend /similar /interact /trending /health How It Works
Catalog Ingestion
POST /ingest-items — product/content descriptions vectorized and indexed.
Interaction Tracking
POST /interact — user views, purchases, and ratings stored persistently for collaborative filtering.
Hybrid Retrieval
Combines vector similarity with interaction history to rank candidates.
AI Ranking
LLM re-ranks candidates and generates natural-language explanations for each recommendation.
Trending Aggregation
Interaction signals aggregated to surface globally trending items.
Use Cases
E-Commerce Product Recs
Show "You might also like" suggestions based on browsing and purchase history.
Content Discovery
Recommend articles, videos, or courses based on what users have consumed.
Similar Item Search
Power "More like this" features using semantic similarity across your catalog.
Trending Dashboard
Surface trending items based on aggregate interaction signals across all users.
Opens Aerostack dashboard to deploy this template
What's Included
Pipeline
Billing Model
metered
Pay per token used. Free tier included.