How a lightweight open-source agent competes with the best in RAG-powered research
What happens when you strip down the architecture of a research agent to its bare essentials? Stefan Webb from Zilliz shares exactly that in this Stack Session presentation on “Deep Searcher” a lightweight, open-source agent designed for teaching and practical experimentation.
Built on Milvus, the most widely adopted open-source vector database, Deep Searcher delivers high-quality results using just semantic search, a reasoning loop, and open models.
Presentation Highlights
This talk is for AI/ML engineers, data scientists, and developers interested in autonomous agents, vector databases, or RAG workflows:
- Why 90% of new data is unstructured and what that means for search
- How Milvus supports scalable, low-latency vector search across billions of items
- Key components of a research agent: subqueries, routing, reflection, and synthesi
- When to use reasoning models, and how to prompt them for iterative decision-making
- How to replicate the entire pipeline with open tools and a single Colab notebook
About The Speaker
Stefan Webb is a Developer Advocate at Zilliz, the company behind Milvus. He helps developers build scalable, vector-powered AI applications, especially in the areas of RAG, semantic search, and generative agents.
Previously a researcher, Stefan now focuses on making advanced ML infrastructure accessible to engineers and practitioners through open-source tools, talks, and code tutorials
3 Days of Context, Insights, & Connections
The 6th annual MLOps World | GenAI Summit is taking place October 7–9, 2025 at the Austin Renaissance Hotel.
Don’t miss this chance to accelerate and de-risk your agentic, GenAI, LLM/SLM, and AI infrastructure projects through cutting-edge strategies, real-world case studies, best practices, and technical deep dives.
Every presentation is hand-picked by a 75 member Steering Committee composed of top AI practitioners whose primary goal is to ensure that their industry colleagues discover the future of AI in production, right now.
The experience also includes a vibrant expo, where attendees shift from focused learning to active participation by engaging in hands-on workshops, Brain Dates, Community Stage, Startup Zone, and interactive demos with leading vendors like Weights & Biases, Outerbounds, and Databricks.
MLOps World | GenAI Summit 2025 is a compact, high-impact way for AI engineers, agentic builders, software engineers, solution architects, infra teams, startups, and enterprise teams to build vital industry contacts and accelerate projects.
Early Bird tickets are on sale now and offer 15% savings when you register in advance.