OpenSearch With Gaia: Building Next-Gen AI-Driven Search and Analytics Workflows
summary
Gaia runs as a local OpenAI‑compatible node serving Llama 3.2 by default and the demo swaps in Quen 3 4B to run on‑device RAG against an OpenSearch vector index (v2.19 binary vectors). The demo loads about 80,000 airline reviews, retrieves ~10 hits per query, converts hits to a prompt, and returns natural‑language insights through a Streamlit UI — the live demo works. Uses Q5 quantization for speed, snapshot embeddings stored in Quadrant or OpenSearch, and supports local‑only mode.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:50Gaia and local AI
- 5:41Demo setup and Gaia node
- 7:49Local inference and OpenSearch
- 11:42RAG pipelines with OpenSearch and Gaia
- 14:52Live demo: airline review insights
- 21:38Customizing models and selection
- 27:18Handling hallucinations and feedback
- 30:09Gaia vs traditional workflows
- 34:49Snapshots, embeddings, and vectors
- 37:16Gaia vs Olama and flexibility
- 39:41Deterministic AI in production
- 41:05Closing and Q&A
related talks
38mA Visual AI Search Flow Builder for OpenSearch: Build GenAI and Vector...
26mOpenSearch: Conversations with Context
39mExperiments with OpenSearch and AI
28mUse OpenSearch as Your Vector Search Database To Optimize GenAI Accuracy!
38mAI-Ready Data Foundation Using Amazon OpenSearch
34mImplementing an open-source RAG with OpenSearch
36mBuilding Next-gen Search With OpenSearch
37mAgentic AI Meets Search: Real-World Use Cases With...