AI-Powered Personalized Discovery with Graph Search and VLM-Enhanced Embeddings
summary
Furniture.com describes a deployed discovery stack: a fine-tuned semantic search hosted on Databricks with VLM-enhanced embeddings that synthesize image and partner data, a Neo4j graph for relational personalization, and real-time ranking that runs at query time. Claims include 2 million products across 77 retailers, 50x faster latency than OpenAI and 13% higher relevance, plus 10+ production models and a personalization pipeline that serves tailored results under 1s.
AI-written summary · about
chapters (6)
- 0:00Furniture.com market challenge
- 2:01Semantic search with VLM embeddings
- 6:42Personalization with user scores
- 10:59Graph personalization with Neo4j
- 17:40Live personalized demo
- 21:47Closing remarks and Q&A
related talks
31mAdvanced search for your e-commerce website
40mVespa.ai’s Personalized Search: Advanced Ranking & Tensor framework
39mAI-Powered Search in Modern E-Commerce
22mMachine Learning: The Engine Behind Modern E-commerce Search
47mBoosting AI With OpenSearch: Vector Stores, Similarity Search, and Retrieval-Augme...
27mMachine Learning for Advanced AI Search
20mBeyond Relevance: Using LLMs to Transform E-commerce Search into Product Discovery Engines
17mEnhancing Chat-Based Customer Recommendations With GraphRAG and Hybrid Search