Putting AI in Production
summary
Claims that Postgres becomes an AI database by combining PGVector for HNSW/IVF similarity search with an EDB Postgres AI extension called AIDB that auto-generates embeddings from S3 or tables, stores vectors, and runs retrieval augmented generation. Presenter cites a GitHub stat of 98% growth in AI apps, demos a Streamlit image search using CLIP and ~132 lines of code, shows a live demo that briefly fails then succeeds, and notes embeddings must be recomputed when switching models.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:12AI hype vs production
- 4:10PostgreSQL for AI workloads
- 5:38LLMs and embeddings
- 8:06Similarity search with PG Vector
- 9:26End-to-end AI apps
- 13:20Recommendation engine use case
- 16:12AIDB simplifies AI pipelines
- 19:28Live demo of AIDB
- 22:23EDB PostgreSQL AI pillars
- 23:28Conclusion
- 24:36Q&A
related talks
44mIntegrating AI with Postgres...
35mPostgres in Artificial Intelligence
34mAnalytics and AI with PostgreSQL - Mission Possible!
34mAnalytics and AI with PostgreSQL - Mission Possible!
37mRound Table Discussion
45mUsing Postgres + OpenAI to power your AI Recommendation...
53mModern PostgreSQL for AI...
18mAccelerate generative AI development with integrated data platforms