AI-Powered Search Book is Now Available on Amazon!
⏳The Wait is Over!
AI-Powered Search was released today (both Print and Kindle) on Amazon! 📕
This essential guide equips you with the tools and insights to build smarter, faster, and more scalable search systems using AI and machine learning. Packed with 200+ code examples, it’s the resource you’ve been waiting for.
🔗 Get your copy on Amazon: https://t.co/EAahXKzDGv
✨ Transform how you think about search. Start your journey with AI-powered relevance today!
#AIPoweredSearch #SemanticSearch #AI #BookLaunch @ManningBooks@treygrainger@softwaredoug@binarymax
Super exciting win for Agentic Search and Late Interaction! 🧬
GPT-5 + Reason-ModernColBERT (150M) reaches ~88% accuracy with an average of ~13 search calls.
For reference, when BrowseComp-Plus was published in August 2025, the max accuracy reported was ~70% using GPT-5 + Qwen3-Embed-8B, using ~22 search calls.
Searching with reasoning 🤖💭is a beast. 🔥
This is a huge evangelist for semantic search and Late Interaction models are particularly shining thanks to their effectiveness at long input modeling with fine-grained similarity scores. 🛠️
Congratulations @antoine_chaffin and team! 🎉
By providing LLMs with 🗺️
•metadata about the collections in the database and their schemas
• values of attributes and their distributions
• initial search results
... We get much better Query Understanding!
This leads to more effective Search, Question Answering, Research, and other AI systems. 🚀
This is an awesome recap of topics from @softwaredoug and my discussion last week with @CShorten30 on the @weaviatepodcast.
We're also diving into these topics deeper in our @aiPoweredSearch course (late enrollment closes tomorrow: https://t.co/2wQhLAEMx0)
Check out the podcast below.
@lateinteraction Yeah, I know people getting crazy offers right now, and not just from the labs.
We're teaching this skillset in our AI Search course to many top engineers (see company logos below). Just kicked off the second cohort (w/ late registration still available)
https://t.co/T5hdbaITWA
I was deep in the newborn trenches🍼as @a1zhang and @lateinteraction took the AI world by storm with RLMs and missed out a bit on these discussions, hoping to catch up! 🧩🔥
Really happy that @softwaredoug brought up RLMs in our discussion of AI-Powered Search, describing how LLMs can write code to search over their context as a Python variable. 👇
♻️ Feedback Loops for Search ♻️
In this clip, @treygrainger explains Reflected Intelligence, a framework for thinking about how to incorporate feedback from users to improve search systems. RL for Search! 🔎 ♻️
Adding a Popularity Feature ✨ to Search Ranking is trickier than you may think.
@softwaredoug lays out the feedback loop problem, popular results get boosted, get more clicks, get boosted further... ♻️
He then introduces two concrete solutions: (1) holdout traffic that doesn't receive the popularity treatment, and (2) active learning that finds "wide-open, gaping holes" in training data. 🧠
WORMHOLE VECTORS! 🔥
In Hybrid Search, searchers such as BM25 and Vector Search run independently, produce separate result sets, and get merged...
Introducing ... 🥁 ... Wormhole Vectors! 🔥
Instead of merging orthogonal results, traverse between the vector spaces. Run a lexical search, pool the matching documents' dense vectors, and search the dense space from that point. 🌀
Hybrid search runs two queries and hopes the right answer was in one of them. Wormhole vectors let you traverse between the spaces instead. 🗺️
"Purple Mattress", is that a color or a brand?
A memorable and fun example of query disambiguation in action. Search the index for "Purple Mattress", aggregate the entities in the results, and ask an LLM to classify which meaning the user intended.
https://t.co/Zvr0iz8a1q
Is grep all that Agents need for search? 🤔
In this clip, Doug explains why simple lexical tools like grep work so well for agents, the transparent input-output relationship lets the agent reason about why results matched, adjust its strategy, and plan its next query. 👇
RAG Inception! 🌀
In this clip, Trey explains RAG for Query Writing. Instead of just taking the original query as input to an LLM query writer, use retrieved information about the data in the database to get even better queries!