For AI to work with us, it needs to understand usI am SUPER EXCITED to publish the 145th episode of the Weaviate Podcast with Alexis Ross, Manya Bansal, and Niloofar Mireshghallah from humans&!What if AI was more human? Well, we are currently on a fast track towards the opposite! AI models are huge making advances in coding, math and are jev-ing their way through classification. But they...
Vector search looks easy, until you hit production scale.I'm super excited to share a new episode of the Weaviate Podcast with Başak from @bookingcom on production-scale vector search, RAG, and agentic AI with @weaviate_io!The podcast begins by discussing Booking's tipping point into adopting vector search and emerging use cases.The scale of Partner-to-Guest messaging alone is insane! There are...
How do we train and evaluate Search Agents?I am SUPER EXCITED to publish a new episode of the Weaviate Podcast with Nandan Thakur on Search Agents!Firstly, congratulations to Nandan who has just completed his Ph.D. at the University of Waterloo advised by Professor Jimmy Lin!During this time he published several impactful works such as BEIR , MIRACL , FreshStack , and many more.This podcast...
Hey everyone! I am SUPER EXCITED to publish a new episode of the Weaviate Podcast with Shreya Shankar on Data Agents!Shreya is a Ph.D. student at UC Berkeley's EPIC Data Lab advised by Aditya Parameswaran. Her research focuses on advancing data systems and human-computer interaction!This podcast dives into her latest work on the Data Agent Benchmark! This is the first benchmark testing how well...
Hey everyone! I am SUPER EXCITED to publish a new episode of the Weaviate Podcast with Amélie Chatelain and Antoine Chaffin on Multi-Vector Search!Amélie, Antoine, and the LightOn team are on fire! They are making breakthrough after breakthrough in Search with Multi-Vector, Late Interaction retrieval models.This podcast covers all sorts of topics, starting with the motivation of Multi-Vector...
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...
In many software applications, Search is the core feature. The search bar is the universal interface for interacting with data!So how should you think about designing user experiences around search so that your users are delighted to use your software?What are the different types of search experiences you should be aware of as someone working in search engineering?Also, whereas LLMs have...
In this clip from the latest Weaviate Podcast, Doug Turnbull 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.https://x.com/weaviatepodcast/status/2028570908978262458
Hey everyone! I am super excited to publish a new episode of the Weaviate Podcast with Doug Turnbull and Trey Grainger on AI-Powered Search!Doug and Trey are both tenured experts in the world of search and relevance engineering. This one is packed with information!Covering designing search experiences, types of search, user interfaces for search, filters, the nuances of agentic search, using...
I am SUPER EXCITED to publish the 132nd episode of the Weaviate Podcast with Thomas van Dongen, head of AI engineering at Springer Nature!Thomas is the creator of Pyversity, a fast, lightweight open-source python library for diversifying retrieval results!Diversity is such an underrated topic in AI and Vector Databases. Whether searching through e-Commerce products or Scientific papers, we...
I am SUPER EXCITED to publish the 132nd episode of the Weaviate Podcast with Thomas van Dongen, head of AI engineering at Springer Nature!Thomas is the creator of Pyversity, a fast, lightweight open-source python library for diversifying retrieval results!Diversity is such an underrated topic in AI and Vector Databases. Whether searching through e-Commerce products or Scientific papers, we...
AI is transforming Database Systems. Perhaps the biggest impact so far has been natural language to query language translations, or Text-to-SQL. However, another massive innovation is brewing.I am SUPER EXCITED to publish the 131st episode of the Weaviate Podcast with Matthew Russo, a Ph.D. student at MIT!AI presents new Semantic Operators for our query languages. For example, we are all...
I am SUPER EXCITED to publish the 130th episode of the Weaviate Podcast featuring Xiaoqiang Lin, a Ph.D. student at the National University of Singapore! During his time at Meta, Xiaoqiang lead the research behind REFRAG: Rethinking RAG-based Decoding!Traditional RAG systems use vectors to find relevant contexts with semantic search, but then throw away these vectors when it is time to pass the...
REFRAG from Meta Superintelligence Labs is a SUPER exciting breakthrough that may spark the second summer of Vector Databases! REFRAG illustrates how Database Systems are becoming even more integral to LLM inference!By making clever use of how context vectors are integrated with LLM decoding, REFRAG is able to make TTFT (Time-to-First-Token) 31X faster and TTIT (Time-to-Iterative-Token) 3X...
I am SUPER excited to publish the 128th episode of the Weaviate Podcast featuring Charles Pierse!Charles has lead the development behind the GA release of Weaviate’s Query Agent!The podcast explores the 6 month journey from alpha release to GA! Starting with the meta from unexpected user feedback, collaboration across teams within Weaviate, and the design of the Python and TypeScript clients.We...
Hey Hacker News community,I think this is a really cool example of prompt optimization with DSPy and the newest GEPA optimizer! I hope you find it interesting!https://www.youtube.com/watch?v=H4o7h6ZbA4o
I am SUPER EXCITED to publish the 127th episode of the Weaviate Podcast featuring Lakshya A. Agrawal!Lakshya is a Ph.D. student at U.C. Berkeley, where he has lead the research behind GEPA: Reflective Prompt Evolution can Outperform Reinforcement Learning!GEPA is a huge step forward for automated prompt optimization, DSPy, and the broader scope of integrating LLMs with optimization...
GEPA is a SUPER exciting advancement for DSPy and a new generation of optimization algorithms re-imagined with LLMs!Starting with the title of the paper, the authors find that Reflective Prompt Evolution can outperform Reinforcement Learning!!Using LLMs to write and refine prompts (for another LLM to complete a task) is outperforming (!!) highly targeted gradient descent updates using...
Multi-Vector Retrieval methods, such as ColBERT and ColPali, offer powerful search capabilities by combining the power of cross encoders and learned representations. This is achieved with the late interaction distance function and contextualized token embeddings. However! The associated costs of storing, indexing, and searching these expanded vector representations is a major challenge!Enter...
Scaling Judge-Time Compute!I am SUPER EXCITED to publish the 121st episode of the Weaviate Podcast featuring Leonard Tang, Co-Founder of Haize Labs!Evals are one of the hottest topics out there for people building AI systems. Leonard is absolutely at the cutting edge of this, and I learned so much from our chat!The podcast covers tons of interesting nuggets around how LLM-as-Judge / Reward...
You have probably heard about how you can connect LLMs to external software tools, in systems mostly referred to as “Agents”. As software developers, we now not only have to think about the Human Experience of interacting and learning to use our tools, but also Agent Experience!I am SUPER EXCITED to publish the 116th episode of the Weaviate Podcast featuring Matt Biilmann from Netlfiy and...
The Arctic Embedding model series from Snowflake has been one of the most impactful open-source text embedding models! In addition to the open model, which has helped a lot of companies kick off their own inference and fine-tuning services (including us at Weaviate), the Snowflake team has also published incredible research breaking down all the components of how to train these models!I am...
JSON Mode has been one of the biggest enablers for working with Large Language Models! JSON mode is even expanding into Multimodal Foundation models! But how exactly is JSON mode achieved?There are generally 3 paths to JSON mode:1. Constrained Generation (such as Outlines)2. Begging the model for a JSON response in the prompt3. A two stage process of generate-then-format (or...
Hey everyone! I am SUPER EXCITED to publish our 101st Weaviate Podcast with Devin Petersohn from Snowflake! Devin has had a remarkable career so far in scaling dataframes from building Modin while at UC Berkeley to then marrying the project with Lux at Ponder, and eventually joining Snowflake! This was one of the most educational conversations of my time hosting the Weaviate Podcast!! Devin...
I am SUPER EXCITED to publish the 100th episode of the Weaviate podcast with Lucas Negritto and Bob van Luijt on Generative UIs!This is an amazing example of AI-native applications and the new generation of software! Rather than producing frontend code, what if the generative model just dynamically renders the application?This reminded me heavily of the GameGAN research from NVIDIA, super cool...
ACORN is a breakthrough in Vector Search with Filtering!! I am BEYOND EXCITED to publish the 99th Weaviate Podcast with Liana Patel and Abdel Rodriguez!This podcast dives into all things ACORN! To set the stage, Vector Search with Filtering is very slow at scale for highly selective filters, such as a filter that composes <5% of the index. There have been a few approaches to combat this such...