Feb 3
Loading video player...
Watch more from .local San Francisco → https://www.youtube.com/playlist?list=PL4RCxklHWZ9s7IrElTzddaZ2w5uupd6TQ Subscribe to MongoDB YouTube→ https://mdb.link/subscribe Learn about Automated Embedding in Vector Search, announced today, a new MongoDB capability that lets you run AI powered semantic search on text in MongoDB with plain natural-language queries, powered by the latest Voyage-4 series of embedding models. If you’re building AI agents, RAG systems, or semantic search, come learn how Atlas removes so you can focus on your application, not your infrastructure. We’ll demo the new index definition and query experience, and show how Auto embedding handles the hardest part of vector search - keeping data and embeddings in sync, model selection during indexing and querying, efficient batching for large datasets, error recovery, and rate limit management during indexing. We’ll walk through how efficient tradeoff between cost and retrieval accuracy can be achieved in vector search, by leveraging different models, Matryoshka Representation Learning dimensions, and various quantization options. 00:00:00 Introduction to Automated Embedding 00:00:29 The Evolution of AI-Native Search Applications 00:02:22 Solving the Complexity of Manual Embedding Pipelines 00:05:21 New Feature: Automated Embedding Search Indexes 00:09:08 Key Benefits: Architecture Simplification & Automation 00:11:59 Technical Deep Dive: Indexing & Querying Architecture 00:17:50 Live Demo: Semantic Search with Movies Dataset 00:21:17 Comparing Model Accuracy and Real-Time Performance 00:24:51 Q&A: Multilingual Support and Future Roadmap Visit Mongodb.com → https://mdb.link/MongoDB Read the MongoDB Blog → https://mdb.link/Blog Read the Developer Blog → https://mdb.link/developerblog