The Future of AI Search
Information Retrieval is evolving rapidly. Late interaction models dominate relevance benchmarks and can now scale in production. Wormhole vectors are redefining hybrid search, new engines are launching to efficiently handle agent scale, & System One / Decision Models are outperforming LLMs at many agentic search tasks. Come learn about the future of AI Search from those implementing it!
Mon Oct 19·3:00 PM UTC
Beyond single-vector search: Late Interaction in 2026
Compressing information into a single embedding forces a model to prioritize what it preserves, before it knows what you’ll search for. Late interaction keeps finer-grained representations, opening up new possibilities across text, code, and visual retrieval, with potential benefits for search agents that rely on finding the right evidence. Learn from Amélie to explore where late interaction helps and what it takes to use it in production.
You'll learn from
Amélie Chatelain
Head of Training & Inference | Senior ML Research Engineer @ LightOn
Trey Grainger
Author, "AI-Powered Search" and Founder @ Searchkernel
Doug Turnbull
Led teams at Shopify, Reddit, Wikipedia
Tue Oct 20·4:00 PM UTC
Scaling Late Interaction in OpenSearch with MUVERA
Late interaction models like ColBERT and ColPali beat single-vector retrieval, but they store hundreds of vectors per document, up to 25x the memory, and many vector indexes can't search them directly. The common shortcut, pooling them into one vector, throws most of the gain away. MUVERA offers a better path. Praveen shares reproducible OpenSearch benchmarks showing which optimizations pay off, which don't, and how to choose for your workload.
You'll learn from
Praveen Mohan Prasad
Data & AI @ AWS | OpenSearch Ambassador
Trey Grainger
Author, "AI-Powered Search" and Founder @ Searchkernel
Doug Turnbull
Led teams at Shopify, Reddit, Wikipedia
Wed Oct 21·3:30 PM UTC
Better Hybrid & Agentic Search with Wormhole Vectors
Today’s hybrid search is lacking. It runs separate lexical/BM25 and semantic/dense vector searches, then just fuses the results. Agents juggle the same searches as separate tools. We can do much better! Wormhole vectors unify query understanding, matching, and ranking across lexical, semantic & behavioral vector spaces. Trey Grainger walks through how to implement them & shares benchmarks beating standard hybrid search. Bring your questions!
You'll learn from
Trey Grainger
Author AI-Powered Search
Doug Turnbull
Co-Author AI Powered Search
Thu Oct 22·3:00 PM UTC
Building topk-embed-v1: frontier multi-vector embeddings
Multi-vector embeddings can improve retrieval quality, but their storage and compute costs often make them impractical at scale. This talk shows how topk-embed-v1 tackles that problem by co-designing the embedding model and retrieval engine, producing highly compressible representations that retain strong retrieval quality with a footprint closer to dense embeddings.
You'll learn from
Marek Galovic
CEO & Co-Founder TopK
Doug Turnbull
AI Search Expert w/ 15 years of experience
Trey Grainger
Author, "AI-Powered Search", Founder @ Searchkernel
Mon Oct 26·4:00 PM UTC
Introducing Luxir: Can we build the fastest search engine?
The original creator of Solr just launched a new open source search engine named Luxir. It's a clean-slate design and implementation... but WHY? Today's most widely used search engines run on the JVM and scale by adding more nodes, but you pay for that inefficiency in the cloud every month. As agents and AI apps multiply query volumes, cost per query matters more than ever. Come hear the project vision and see early benchmarks from Yonik himself!
You'll learn from
Yonik Seeley
Solr Creator | Building Luxir
Trey Grainger
Author, "AI-Powered Search" and Founder @ Searchkernel
Doug Turnbull
Led teams at Shopify, Reddit, Wikipedia
Wed Oct 28·3:30 PM UTC
Can a Decision Model Think Like a Search Engineer?
Every search index hides judgment calls. For example, should a name field match misspellings by similar sound? Enable this on the wrong field and you add bad matches. But new decision models return typed answers with calibrated confidence, so can they make those calls? David Tippett, founder at Relevan, tested Jev, Laya, & Clef. It turns out the answer is yes!... in many cases. Hear experiments and benchmarks firsthand, and bring your questions.
You'll learn from
David Tippett
Founder @ Relevan | Search @ Firecrawl
Trey Grainger
Author AI Powered Search
Doug Turnbull
Co-Author AI Powered Search
Thu Oct 29·5:00 PM UTC
Agent Memory with Turbopuffer
We want to start an AI session and pick up where we left off. With products like OpenAI's Dots or personal agents like Hermes / OpenClaw, retrieving past context automatically helps us avoid endless re-explaining and documenting. Can we build a search system that remembers our preferences, how we like to work, and our priorities? We'll explore how anyone can build agentic memory for their own coding agent with Turbopuffer.
You'll learn from
Doug Turnbull
Co-Author AI Powered Search
Trey Grainger
Author AI-Powered Search
Fri Oct 30·3:00 PM UTC
The unreasonable effectiveness of BM25 for agentic search
Grep became a core retrieval tool for coding agents. Jo Kristian Bergum, CEO of Hornet, argues that BM25 serves a similarly powerful function for search agents (in contrast with popular vector embedding approaches). GPT-5 is a very capable searcher, writing long, precise queries with exact phrases in ~98% of sessions. Jo will walk through the unreasonable effectiveness of BM25 for agentic search, and why this 30-year old algorithm often wins.
You'll learn from
Jo Kristian Bergum
CEO at HORNET.dev
Trey Grainger
Author AI Powered Search
Doug Turnbull
Co-Author AI Powered Search
Mon Nov 2·5:00 PM UTC
The AI Search Frontier in 2026
Information Retrieval is evolving at an rapid pace. In our month-long "AI-Powered Search: Modern Retrieval for Humans & Agents" course, we work hard to cover both the tried-and-true AI Search fundamentals and bring additional experts as guest instructors on emerging / frontier topics. This session is a survey of the latter - emerging topics in AI Search that you should know about. Join us to hear the latest advancements this year in AI Search.
You'll learn from
Trey Grainger
Author, AI-Powered Search
Doug Turnbull
Co-author of AI-Powered Search