Late chunking improves context recall in RAG pipelines Weaviate vector database 2026-09-23 | Late chunking improves context recall in RAG pipelines
Weaviate TECH Hands-On: Personalization Agent Weaviate • Vector Database 2025-07-11 | In this hands-on workshop session JP dives into building personalized recommendation systems and user personalized queries for the Weaviate vector database using the Weaviate Personalization Agent.00:00:00 Intro00:02:14 Environment Setup00:12:35 Intro to Personalization Agent00:17:38 Hands-On part: How to use the Personalization Agent00:50:07 Summary and further resources▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit [weaviate.io](http://weaviate.io/)- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Agentic Topic Modeling with Maarten Grootendorst - Weaviate Podcast #126! Weaviate • Vector Database 2025-07-09 | Maarten Grootendorst is a psychologist turned AI engineer who has created BERTopic and authored "Hands-On Large Language Models" with Jay Alammar. The rise of LLMs and Agents are transforming many areas of software! This podcast dives deep into their impact on Topic Modeling! Maarten designed BERTopic from the start with modularity in mind -- letting you ablate embedding models, dimensionality reduction, clustering algorithms, and more. This early insight to prioritize modularity makes BERTopic incredibly well structured to become more "Agentic". An "Agentic" Topic Modeling algorithm can use LLMs to generate topics or topic descriptions, as well as contrast them with other topics. It can decide which topics to subdivide, and it can integrate human feedback and evaluate topics in novel ways... I hope you find the podcast interesting!Links:Hands-On Large Language Models: oreilly.com/library/view/hands-on-large-language/9781098150952BERTopic: github.com/MaartenGr/BERTopicBERTopic (paper): arxiv.org/abs/2203.05794Learn more about Maarten Grootendorst: maartengrootendorst.comTopicGPT: arxiv.org/abs/2311.01449TnT-LLM: arxiv.org/abs/2403.12173Chapters0:00 Welcome Maarten!1:57 Hands-On Large Language Models7:34 An Overview of Topic Modeling10:45 LLM Topic Generation17:13 Topic Modeling with Human Feedback21:33 Topic Granularity26:00 Visualizing Topics31:18 Contrastive Topics33:24 LLM-as-Judge for Topics39:06 Separating Generation from Assignment44:20 Applications of Topic Modeling55:14 Semi-Supervised BERTopic1:01:06 Future Directions for AI
Weaviate TECH Hands-On: Query Agent in JavaScript Weaviate • Vector Database 2025-07-07 | In this session, Daniel walks you through our Weaviate Query Agent and uses the agent to build an Agentic RAG system in JavaScript00:00:00 Intro & Overview00:01:15 What is the Weaviate Query Agent00:07:37 Live Coding with Weaviate Query Agent00:21:00 Further Resources▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Sufficient Context with Hailey Joren - Weaviate Podcast #125! Weaviate • Vector Database 2025-07-02 | Hailey Joren is a Ph.D. student at UCSD! Hailey and collaborators at Duke University and Google have recently published Sufficient Context: A New Lens on Retrieval Augmented Generation Systems in ICLR 2025! There are so many interesting nuggets to this work! Firstly, it really helped me understand the difference between *relevant* search results and sufficient context for answering the question. Armed with this lens of looking at retrieved context, Hailey and collaborators make all sorts of interesting observations about the current state of Hallucination. RAG unfortunately makes the models far less likely to hallucinate, and the existing RAG benchmarks unfortunately do not emphasize retrieval adaptation well enough -- indicated by LLMs outputting correct answers despite insufficient context 35-62% of the time! However, reason for optimism! Hailey and team develop an autorater that can detect insufficient context 93% of the time! There are all sorts of interesting ideas around this paper! I really hope you find the podcast useful!Links:Sufficient Context: A New Lens on Retrieval Augmented Generation Systems - arxiv.org/pdf/2411.06037Chapters0:00 Welcome Hailey1:40 Definition of Sufficient Context4:47 Context Engineering8:42 AutoRater13:08 Self-Rated Confidence23:45 RAG and Long Context LLMs29:35 Search Relevance vs. Sufficient Context35:00 Retrieval-Aware Fine-Tuning44:05 Directions for the future of AI
RAG Benchmarks with Nandan Thakur - Weaviate Podcast #124! Weaviate • Vector Database 2025-06-25 | Nandan Thakur is a Ph.D. student at the University of Waterloo! Nandan has worked on many of the most impactful works in Retrieval-Augmented Generation (RAG) and Information Retrieval. His work ranges from benchmarks such as BEIR, MIRACLE, TREC, and FreshStack, to improving the training of embedding models and re-rankings, and more!Links:Nandan Thakur: thakur-nandan.github.ioFreshStack Benchmarks: fresh-stack.github.ioBEIR Benchmarks: arxiv.org/abs/2104.08663BrowseComp: arxiv.org/abs/2504.12516Sufficient Context: arxiv.org/abs/2411.06037Check out the Weaviate Query Agent! weaviate.io/developers/agents/queryChapters0:00 Welcome Nandan!1:15 The BEIR Benchmarks8:25 Evolution of RAG Benchmarks12:25 Diversity in Search Results18:20 Reasoning and Query Writing23:03 Search Result Summarization26:10 Looping Searches29:45 High Recall Search33:10 Paginating Search Results38:05 Searching with Filters44:35 Mixture of Retrievers51:08 Creating FreshStack Weaviate1:00:35 Exciting Directions for AI
RAG is the Backbone of Enterprise AI Weaviate • Vector Database 2025-06-20 | “You don’t need RAG, context windows are huge now…”Sounds great, right?But the reality is: 𝗯𝗶𝗴𝗴𝗲𝗿 𝗶𝘀𝗻’𝘁 𝗮𝗹𝘄𝗮𝘆𝘀 𝗯𝗲𝘁𝘁𝗲𝗿, especially if it’s slower, costlier, and oblivious to updates.Join Box’s CTO Ben Kus, along with Bob van Lujit and Connor Shorten, in the latest Weaviate Podcast to learn more about enterprise-scale AI solutions, Box's three-layer infrastructure puzzle, embeddings, retrieval, and much more!➡️ youtube.com/watch?v=pPvSur8iEXY
Weaviate 1.31 Release Highlights & More Weaviate • Vector Database 2025-06-18 | In this session JP and Marion walk through some of the highlights of Weaviate 1.31.00:00:00 Intro & Overview00:03:00 MUVERA00:15:50 Flexible Vectorizer Changes00:21:00 Shard Movements - comming in 1.3200:27:00 Weaviate Cloud - Import Data Tool00:44:30 Shard Movement - Replication Factor00:50:30 HNSW Snapshots▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit: http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Generic Search is Dead. Weaviate • Vector Database 2025-06-04 | After introducing the Query Agent and the Transformation Agent, it's time to meet the Personalization Agent - search that adapts to each user.This isn't just another ranking tool. It's the first step toward truly personalized search that understands:• User profiles (preferences, interests, characteristics)• Real-time interactions (likes, dislikes, engagement patterns)• LLM reasoning included: Custom instructions for specific use casesWhether you're building a clothing brand or recipe platform (recipes below 🧑🍳), imagine search results that actually match what each user wants to see.The coolest part? You can choose classic ML or go full LLM with agent ranking that even explains its reasoning.Blog + Recipes: weaviate.io/blog/personalization-agent?utm_source=channels&utm_medium=w_social&utm_campaign=agents&utm_content=video_post_268065893Tuana Socials- Twitter/X: https://x.com/tuanacelik- LinkedIn: linkedin.com/in/tuanacelikFemke Socials- Twitter/X: https://x.com/femke_plantinga- LinkedIn: linkedin.com/in/femke-plantinga- TikTok: tiktok.com/@femkeplant.ai- Instagram: instagram.com/femkeplant.ai▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit [weaviate.io](http://weaviate.io/)- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
MUVERA with Rajesh Jayaram and Roberto Esposito - Weaviate Podcast #123! Weaviate • Vector Database 2025-05-28 | Multi-vector retrieval offers richer, more nuanced search, but often comes with a significant cost in storage and computational overhead. How can we harness the power of multi-vector representations without breaking the bank? Rajesh Jayaram, the first author of the groundbreaking MUVERA algorithm from Google, and Roberto Esposito from Weaviate, who spearheaded its implementation, reveal how MUVERA tackles this critical challenge.Dive deep into MUVERA, a novel compression technique specifically designed for multi-vector retrieval. Rajesh and Roberto explain how it leverages contextualized token embeddings and innovative fixed dimensional encodings to dramatically reduce storage requirements while maintaining high retrieval accuracy. Discover the intricacies of quantization within MUVERA, the interpretability benefits of this approach, and how LSH clustering can play a role in topic modeling with these compressed representations.This conversation explores the core mechanics of efficient multi-vector retrieval, the challenges of benchmarking these advanced systems, and the evolving landscape of vector database schemas designed to handle such complex data. Rajesh and Roberto also share their insights on the future directions in artificial intelligence where efficient, high-dimensional data representation is paramount.Whether you're an AI researcher grappling with the scalability of vector search, an engineer building advanced retrieval systems, or fascinated by the cutting edge of information retrieval and AI frameworks, this episode delivers unparalleled insights directly from the source. You'll gain a fundamental understanding of MUVERA, practical considerations for its application in making multi-vector retrieval feasible, and a clear view of future directions in AI.Links:MUVERA: arxiv.org/abs/2405.19504CRISP: arxiv.org/pdf/2505.11471ColBERT: arxiv.org/abs/2004.12832ColPali: arxiv.org/abs/2407.01449Multi-Vector Embeddings with Weaviate (Tutorial): weaviate.io/developers/weaviate/tutorials/multi-vector-embeddingsChapters0:00 Welcome Rajesh and Roberto2:10 Intro to Multi-Vector Retrieval7:53 Contextualized Token Embeddings14:04 Interpretability of Multi-Vector Retrieval17:46 Multi-Vector Storage Cost20:10 MUVERA Deep Dive32:30 Fixed Dimensional Encodings55:32 Quantization in MUVERA1:00:04 LSH Clustering for Topic Modeling1:02:44 Benchmarks for Multi-Vector Retrieval1:06:33 Future of Vector Database Schemas1:09:28 Directions for the Future of AI
Patronus AI with Anand Kannappan - Weaviate Podcast #122! Weaviate • Vector Database 2025-05-15 | AI agents are getting more complex and harder to debug. How do you know what's happening when your agent makes 20+ function calls? What if you have a Multi-Agent System orchestrating several Agents? Anand Kannappan, co-founder of Patronus AI, reveals how their groundbreaking tool Percival transforms agent debugging and evaluation. Percival can instantly analyze complex agent traces, it pinpoints failures across 60 different modes, and it automatically suggests prompt fixes to improve performance. Anand unpacks several of these common failure modes. This includes the critical challenges of "context explosion" where agents process millions of tokens. He also explains domain adaptation for specific use cases, and the complex challenge of multi-agent orchestration. The paradigm of AI Evals is shifting from static evaluation to dynamic oversight! Also learn how Percival's memory architecture leverages both episodic and semantic knowledge with Weaviate!This conversation explores powerful concepts like process vs. outcome rewards and LLM-as-judge approaches. Anand shares his vision for "agentic supervision" where equally capable AI systems provide oversight for complex agent workflows. Whether you're building AI agents, evaluating LLM systems, or interested in how debugging autonomous systems will evolve, this episode delivers concrete techniques. You'll gain philosophical insights on evaluation and a roadmap for how evaluation must transform to keep pace with increasingly autonomous AI systems.Links:Percival Launch: patronus.ai/percivalDocs: docs.patronus.ai/docs/percivalPaper: arxiv.org/abs/2505.08638Chapters0:00 Welcome Anand!1:15 Percival!17:20 Online and Offline Agent Tracing20:40 Complex Agent Traces23:05 Quick Insights and Deep Research24:47 Automated Agent Tuning31:19 LLM-as-Judge and Scalable Oversight42:24 Agent Inbox for Evals45:49 Causal Inference and AI51:24 Percival and Weaviate56:04 Exciting Directions for AI
Embedding model evaluation & selection guide Weaviate • Vector Database 2025-05-14 | Selecting the right embedding model can make or break your AI application's performance. In this guide, JP from Weaviate walks you through a practical 4-step framework to navigate the complex landscape of embedding models.Learn how embedding models transform content into numerical vectors that power search, recommendations, and more. Discover why your model choice impacts not just performance, but also resource requirements and operational costs.This video is a summary of our in-depth guide on Weaviate Academy. You can find the guide here: weaviate.io/developers/academy/theory/embedding_model_selection?utm_source=youtube&utm_medium=w_social&utm_campaign=dev_education&utm_content=video_post_680568036⏱️ TIMESTAMPS:0:00 The cookie recipe challenge0:41 Introduction1:05 Why embedding model selection matters2:10 The 4-step selection framework4:40 Practical tips for better decision-making5:36 Recap & resources🔑 KEY TAKEAWAYS:- How to identify your specific needs across data characteristics, performance requirements, operational factors, and business constraints- Strategies for narrowing down hundreds of models to a manageable shortlist- Methods for rigorously evaluating models using both standard benchmarks and your own data- When and how to reassess your model selection as the field evolves👨💻 RESOURCES:In-depth guidance, code examples, and evaluation metrics, visit our comprehensive guide at Weaviate Academy: weaviate.io/developers/academy/theory/embedding_model_selectionConnect with JP on LinkedIn: linkedin.com/in/jphwangMTEB leaderboard: huggingface.co/spaces/mteb/leaderboard▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub: github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a questions?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
The Transformation Agent: Natural Language Database Transformations Weaviate • Vector Database 2025-05-13 | Managing and transforming data in your database can feel like a never-ending task...But what if we told you could simply tell your database what changes you want to make to your data, in plain language, and have it handle all the technical details for you.That's exactly where Weaviate's new Transformation Agent comes in!The Transformation Agent lets you modify and enhance your data using natural language instructions. It can:- Add new properties to objects based on existing data- Update existing properties with enhanced information- Process multiple operations in parallel- Apply transformations across entire collectionsThis agent handles the tedious task of designing database updates and lets you focus on what information you need, not how to get it.Blog + Recipe: weaviate.io/blog/transformation-agentTuana Socials- Twitter/X: https://x.com/tuanacelik- LinkedIn: linkedin.com/in/tuanacelikFemke Socials- Twitter/X: https://x.com/femke_plantinga- LinkedIn: linkedin.com/in/femke-plantinga- TikTok: tiktok.com/@femkeplant.ai- Instagram: instagram.com/femkeplant.ai▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit: http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Haize Labs with Leonard Tang - Weaviate Podcast #121! Weaviate • Vector Database 2025-05-12 | How do you ensure your AI systems actually do what you expect them to do? Leonard Tang takes us deep into the revolutionary world of AI evaluation with concrete techniques you can apply today. Learn how Haize Labs is transforming AI testing through "scaling judge-time compute" - stacking weaker models to effectively evaluate stronger ones. Leonard unpacks the game-changing Verdict library that outperforms frontier models by 10-20% while dramatically reducing costs. Discover practical insights on creating contrastive evaluation sets that extract maximum signal from human feedback, implementing debate-based judging systems, and building custom reward models that align with enterprise needs. The conversation reveals powerful nuggets like using randomized agent debates to achieve consensus and lightweight guardrail models that run alongside inference. Whether you're developing AI applications or simply fascinated by how we'll ensure increasingly powerful AI systems perform as expected, this episode delivers immediate value with techniques you can implement right away, philosophical perspectives on AI safety, and a glimpse into the future of evaluation that will fundamentally shape how AI evolves.Learn more about Haize Labs! - haizelabs.comCheck out Verdict on GitHub - github.com/haizelabs/verdictChapters0:00 Weaviate Podcast #121!0:46 Welcome Leonard!1:16 Founding Haize Labs8:31 UX for Evals17:01 Scaling Judge-Time Compute with Verdict23:26 Debate Judges26:06 Compute Scaling28:50 Declarative Judge Pipelines31:13 Custom Reward Models37:20 Reasoning in Reward Models39:20 Mechanistic Interpretability45:30 Guardrails in Inference Pipelines47:35 Can we control Superintelligence?52:08 Exciting Directions for AI
Box AI with Ben Kus and Bob van Luijt - Weaviate Podcast #120! Weaviate • Vector Database 2025-05-07 | Ben walks us through Box's three-layer infrastructure puzzle: First, the mind-boggling base infrastructure (think millions of interactions per second and trillions of files). Second, their unique multi-tenant security challenge - unlike most SaaS platforms, Box users share content across company boundaries, making traditional tenant isolation impossible. And third, ensuring AI respects all these complex permissions while still delivering value.The podcast then dives further into how vector embeddings can balloon file sizes - a few hundred bytes of text can require 4-6KB of vector data storage! We also dig into why RAG remains essential despite growing context windows, and how Box is developing AI agents that transform painful enterprise processes like RFP responses.A fun, insightful look at what happens when cutting-edge AI meets enterprise-grade security and scale requirements!Chapters0:00 Weaviate Podcast #120!0:20 Welcome Ben!1:10 Exabyte Scale at Box8:10 Infrastructure Rent vs. Buy12:30 Founder-Led AI Adoption19:37 Embeddings, RAG, and The AI Tipping Point28:05 How many Vectors per Data Object?37:20 Storage Tiers and Dynamic Indices in Retrieval45:00 Agents51:15 Agents for RFPs
Reimagine Data Workflows with Weaviate Agents Weaviate • Vector Database 2025-04-30 | Calling all AI devs, tech leads, and novices experimenting with agentic AI!Join us for a hands-on walkthrough of Weaviate Agents — a powerful suite of services designed to simplify and automate your AI data workflows.In this live session, we'll showcase how the 𝐐𝐮𝐞𝐫𝐲 𝐀𝐠𝐞𝐧𝐭, 𝐓𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧 𝐀𝐠𝐞𝐧𝐭, 𝐚𝐧𝐝 𝐏𝐞𝐫𝐬𝐨𝐧𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐀𝐠𝐞𝐧𝐭 work under the hood to enable natural language querying, real-time data transformation, and context-aware personalization without heavy lifting.This session will include live demos, example use cases, and implementation tips.[𝐈𝐦𝐩𝐨𝐫𝐭𝐚𝐧𝐭] 𝐈𝐟 𝐲𝐨𝐮 𝐥𝐢𝐤𝐞 𝐭𝐨 𝐫𝐞𝐜𝐞𝐢𝐯𝐞 𝐦𝐚𝐭𝐞𝐫𝐢𝐚𝐥𝐬 𝐮𝐬𝐞𝐝 𝐝𝐮𝐫𝐢𝐧𝐠 𝐭𝐡𝐞 𝐬𝐞𝐬𝐬𝐢𝐨𝐧 𝐚𝐧𝐝 𝐟𝐨𝐥𝐥𝐨𝐰-𝐮𝐩 𝐢𝐧𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧 𝐦𝐚𝐤𝐞 𝐬𝐮𝐫𝐞 𝐭𝐨 𝐚𝐠𝐫𝐞𝐞 𝐭𝐨 𝐖𝐞𝐚𝐯𝐢𝐚𝐭𝐞𝐬 𝐏𝐫𝐢𝐯𝐚𝐜𝐲 𝐏𝐨𝐥𝐢𝐜𝐲
Stateful Agents self-optimize their Context Window Weaviate • Vector Database 2025-04-25 | Have you tried asking ChatGPT what it knows about you? This clip from Sarah Wooders describes how Letta is pioneering a new wave of Stateful Agents that self-optimize their context windows based on your message history and external documents. Since recording this podcast, Letta has continued to push the frontier of Stateful Agents with the introduction of "Sleep-time Compute"! I think it is well worth your time to take a look at their announcement resources below, and I also hope this clip inspires your interest in the podcast with Sarah!letta.com/blog/sleep-time-compute
The Query Agent Weaviate • Vector Database 2025-04-24 | Navigating multiple databases can feel like searching through a sea of information.(But guess what, you don’t have to anymore!)We’re introducing the Query Agent - an AI assistant that seamlessly searches across your Weaviate collections!Using the power of LLMs, Query Agent automatically:- Decides which collections to search- Combines information from multiple sources- Provides comprehensive answers to complex questions✨ Available now for all Weaviate Cloud & free sandbox users!Try it today: weaviate.io/developers/wcs?utm_source=channels&utm_medium=w_social&utm_campaign=agents&utm_content=video_post_68040974500:00 Introduction: The Query Agent00:25 Example Clothing Brand (Femke)01:35 Query Agent Demo (Tuana)03:00 How to Try The Query AgentTuana Socials- Twitter/X: https://x.com/tuanacelik- LinkedIn: linkedin.com/in/tuanacelikFemke Socials- Twitter/X: https://x.com/femke_plantinga- LinkedIn: linkedin.com/in/femke-plantinga- TikTok: tiktok.com/@femkeplant.ai- Instagram: instagram.com/femkeplant.ai▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit [weaviate.io](http://weaviate.io/)- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
UX? DX? Meet AX! Weaviate • Vector Database 2025-04-22 | Too agent-friendly = humans get lost.Too human-friendly = agents break.So… who are you really designing for?Join Matt Biilmann, Sebastian Witalec, Charles Pierse, and Connor Shorten on the Weaviate Podcast to learn all about Agent Experience (AX) - and why it’s becoming so important alongside the user and developer experience.
10. Use Cases of AI Agents Weaviate • Vector Database 2025-04-14 | The future of work isn’t human OR AI.It’s human WITH AI Agents. Let’s have a look at the industries that AI is transforming right NOW:
9. The Future of AI Agents Weaviate • Vector Database 2025-04-11 | How much power should we give AI Agents? Here’s what the future really looks like:
AI Agents Explained: Making AI Actually WORK For You Weaviate • Vector Database 2025-04-10 | How can you make AI agents work for you? At Weaviate, we've developed three specialized agents to make working with your data easier:1️⃣ Query Agent: Finds exactly what you need from your database when you ask in natural language2️⃣ Transformation Agent: Lets you define database operations conversationally (e.g., "create a summary of this text")3️⃣ Personalization Agent (coming soon): Gives smarter recommendations by incorporating user behavior and preferences🔔 Stay TUNED for in-depth videos on each Weaviate Agent + Demos 🔔TIMESTAMPS00:00 Introduction: Beyond Basic Language Models00:17 What Is an AI Agent?00:35 How do AI Agents Actually Work?01:27 Memory & Context Retention01:45 How Can We Build AI Agents With Weaviate?01:50 Query Agent01:57 Transformation Agent02:10 Personalization Agent 02:28 How to Try Weaviate Agents TodayTuana Socials- Twitter/X: https://x.com/tuanacelik- LinkedIn: linkedin.com/in/tuanacelikFemke Socials- Twitter/X: https://x.com/femke_plantinga- LinkedIn: linkedin.com/in/femke-plantinga- TikTok: tiktok.com/@femkeplant.ai- Instagram: instagram.com/femkeplant.ai▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit [weaviate.io](http://weaviate.io/)- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Weaviate 1.28 Release Highlights Weaviate • Vector Database 2025-04-10 | In this session JP walks through some of the highlights of Weaviate 1.28.00:00:00 Intro00:01:30 RBAC Overview & Demo00:36:10 Async Indexing00:40:00 Conflict resolution for Multi-node clusters00:43:18 Japanese Tokenizer00:45:55 BlockMax WAND implementation for BM2500:48:07 Voyage AI Multimodal Model00:49:50 Weaviate embeddings▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Weaviate Tech Hands-On: Query Agent Weaviate • Vector Database 2025-04-10 | In this session Tuana demonstrates in a hands-on technical walkthrough how to use the new Weaviate Query Agent. She uses natural language in order to query data spread across multiple collections and derive meaningfull insights from it while the Query Agent handles the search logic for her.00:00:00 Intro00:01:40 What is the Weaviate Query Agent00:04:30 Search vs. Aggregation operations00:05:17 setup eCommerce Dataset in Weaviate00:10:40 Setup the Weaviate Query Agent00:11:30 Run the Weaviate Query Agent00:21:30 Run context-aware follow-up questions00:25:59 Search across multiple collections00:40:20 Change system prompts for the Weaviate Query Agent▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit [weaviate.io](http://weaviate.io/)- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud Services for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-ioTuana Socials- Twitter/X: https://x.com/tuanacelik- LinkedIn: linkedin.com/in/tuanacelik
What is Vibe Coding? Weaviate • Vector Database 2025-04-10 | Vibe coding is shifting the development focus from specific implementation details and knowledge to overall architecture and wisdom. Domain knowledge, architectural thinking, and system design principles provided by humans will still be necessary, but just expressed through prompts and edits rather than direct implementation. Check out Etienne’s experiment: https://x.com/etiennedi/status/1899843351030964535
Structured Outputs with Will Kurt and Cameron Pfiffer - Weaviate Podcast #119! Weaviate • Vector Database 2025-04-09 | Hey everyone! Thanks so much for watching another episode of the Weaviate Podcast! Dive into the fascinating world of structured outputs with Will Kurt and Cameron Pfeiffer, the brilliant minds behind Outlines, the revolutionary open-source library from .txt.ai that's changing how we interact with LLMs. In this episode, we explore how constrained decoding enables predictable, reliable outputs from language models—unlocking everything from perfect JSON generation to guided reasoning processes.Will and Cameron share their journey to founding .txt.ai, explain the technical magic behind Outlines (hint: it involves finite state machines!), and debunk misconceptions around structured generation performance. You'll discover practical applications like knowledge graph construction, metadata extraction, and report generation that simply weren't possible before this technology.Whether you're building AI systems or curious about where the field is heading, you'll gain valuable insights on how structured outputs integrate with inference engines like vLLM, why multi-task inference outperforms single-task approaches, and how this technology enables scalable agent systems that could transform software architecture forever. Join us for this mind-expanding conversation about one of AI's most important but underappreciated innovations—and discover why the future might belong to systems that combine freedom with structure.Links:dottxt AI: dottxt.coMaking LLMs Reliable: Building an LLM-powered Web App to Generate Gift Ideas: blog.dottxt.co/gifter.htmlSay What You Mean: A Response to 'Let Me Speak Freely': blog.dottxt.co/say-what-you-mean.htmlCoalescence: making LLM inference 5x faster: blog.dottxt.co/coalescence.htmlStructuredRAG: JSON Response Formatting with Large Language Models: arxiv.org/abs/2408.11061Chapters0:00 Welcome Will and Cameron!1:42 What lead you to dottxt?7:20 Structured Outputs for Beginners9:00 Metadata Extraction17:36 Structured Reasoning23:00 Report Generation28:25 Multi-Task Inference30:55 How does Outlines work?35:55 Integration with vLLM and Inference Engines43:55 Let Me Speak Freely: A Rebuttal56:35 Distribution Alignment1:04:10 Exciting Directions for AI
8. AI Agents: Tools & APIs Weaviate • Vector Database 2025-04-09 | An AI agent's power lies in the tools it can use!Tools and APIs enable agents to:
What is MCP (Model Context Protocol)? Weaviate • Vector Database 2025-04-08 | 𝗠𝗼𝗱𝗲𝗹 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹 (𝗠𝗖𝗣) is turning into the universal connector for AI systems.Think of MCP as the USB-C cable of AI integrations. Instead of building separate connections for each data source or tool, you create one MCP server that exposes specific capabilities (like searching a database or querying weather data).Check out the Weaviate MCP: github.com/weaviate/mcp-server-weaviate
7. AI Agents Frameworks Weaviate • Vector Database 2025-04-07 | Stop building AI agents from scratch. These 6 frameworks will save you hundreds of development hours.More in this series: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680256845
6. AI Agents Explained: Memory & Storage Weaviate • Vector Database 2025-04-04 | How do AI Agents remember your conversations? Let’s break it down in 75 seconds!Watch more videos in this series: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680318483
5. AI Agents Explained: LLMs Weaviate • Vector Database 2025-04-02 | Large language models can do jaw-dropping things.But for AI agents, they need to include these three essential features:More about AI Agents: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680283937
4. “Agentic” AI or AI “Agents”? Weaviate • Vector Database 2025-03-31 | Agentic AI or AI Agents?These buzzwords get mixed up a lot, but they're actually quite different!See the other videos of this series: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680531700
3. The History of AI Agents Weaviate • Vector Database 2025-03-27 | AI agents have technically been around for a long time. (longer than you might think!)Full playlist AI Agents Explained: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680296508
2. The Building Blocks of AI Agents Weaviate • Vector Database 2025-03-27 | How do AI Agents work?AI agents rely on these key building blocks:AI Agents Explained series: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680496173
1. An Introduction to AI Agents Weaviate • Vector Database 2025-03-27 | AI Agents - a key topic in artificial intelligence that's gaining momentum.(They have actually been around for quite some time)In this 10-part series, Femke shares AI Agent fundamentals.Full series: youtube.com/playlist?list=PLTL2JUbrY6tXYK3n_bIYZWr52l1qBITBEDownload the FREE Ebook: weaviate.io/ebooks/agentic-architectures?utm_source=youtube&utm_medium=w_social&utm_campaign=agents&utm_content=videoshort_post_680656377
Building an AI chatbot without any code Weaviate • Vector Database 2025-03-27 | Stop trying to find internal company info like this..This no-code AI solution takes 5 minutes to build.Sometimes it pays to work smarter, not harder - watch Victoira build this company knowledge Q&A system with zero coding, using Stack AI's Platform. Check it this webinar to learn more! events.weaviate.io/agents-stackai-webinar-2025?utm_source=linkedin&utm_medium=vs_social&utm_campaign=agents&utm_content=videoshort_post_680821688
Weaviate TECH Hands-On: Transformation Agent Weaviate • Vector Database 2025-03-27 | In this technical hands-on session we dive into the Weaviate Transformation Agent. With the help of the agent we transform data objects inside Weaviate through natural language. We add new properties and update existing ones to make the data fit our needs. As a bonus we will use the Weaviate Query agent to perform various complex search operations.Topics:00:00 Intro00:22 What is the Weaviate Transformation Agent05:00 Setup a test dataset inside Weaviate15:10 Challenges when working with data19:00 Using agents to add properties26:50 Using agents to perform multiple operations simultaniously33:30 Using agents to update existing properties41:35 Bonus: Using the Weaviate Query Agent44:30 Bonus: Follow Up Queries with the Query Agent47:00 Bonus: Current limitations51:20 Additional Resources▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Speeding Up AI Quantization Techniques for Models and Vector DBs Weaviate • Vector Database 2025-03-26 | In this talk, Marcin Antas (linkedin.com/in/antasmarcin), a senior Core Engineer who's been at @Weaviate for over 4 years, breaks down the essential techniques for optimizing AI models through quantization.Learn how to significantly reduce the memory footprint of large language models and embedding models while preserving their functionality - even on constrained devices like Raspberry Pi 5!🔑 Key Topics Covered:- LLM quantization techniques (from FP16/FP8 to 4-bit precision)- The GGUF format and LLAMA.cpp framework- Why feed-forward layer parameters are more sensitive than attention layers- Embedding model quantization using ONNX- Vector database quantization methods (Product, Binary, and Scalar)- Running vector databases and AI models on edge devicesThis technical deep dive is perfect for developers looking to optimize AI models for memory-constrained environments or deploy vector search capabilities on edge devices.Learn more from Weaviate at https://weaviate.io.
Build a No-Code Agentic Workflow in Under 5 Minutes Weaviate • Vector Database 2025-03-25 | Developers often spend weeks building agentic workflows.Femke just built one in 5 minutes, without writing a single line of code.• Route general questions to Google search• Direct document-specific queries to your knowledge baseUsing Stack AI's no-code platform, I built a workflow that automatically chooses between these routes based on the query type.From searching the web to processing 200+ page documents, building secure, robust, customizable AI applications and pipelines on your own data is way easier than most companies think it is.And of course, powered by Weaviate's vector database for efficient document processing and search 💚📽️ Watch the recording: "How AI Agents Can Transform Your Business" and see a live demo of building intelligent routing agents with Stack AI: events.weaviate.io/agents-stackai-webinar-2025?utm_source=linkedin&utm_medium=fp_social&utm_campaign=stack_ai&utm_content=video_post_806872823Learn more about Weaviate: weaviate.ioExplore StackAI: StackAI: stack-ai.com
Synthetic Data with David Berenstein and Ben Burtenshaw - Weaviate Podcast #118! Weaviate • Vector Database 2025-03-25 | Synthetic Data: The Building Bocks of AI's Future!Hey everyone! I am SUPER EXCITED to publish the 118th episode of the Weaviate Podcast featuring David Berenstein and Ben Burtenshaw from HuggingFace! This podcast explores the intricacies of synthetic data generation, detailing methodologies such as data augmentation, distillation, and instruction refinement. The conversation delves into persona-driven synthetic data, highlighting applications like Persona Hub, and discusses algorithms to enhance diversity, complexity, and quality of generated data. Additionally, they cover integration with Hugging Face’s ecosystem, including Argilla for annotation, AutoTrain for fine-tuning, and advanced data exploration tools like the Data Studio and SQL console. The podcast also touches upon the potential for synthetic image data generation and the exciting future of AI education and accessibility.Further ResourcesArgilla: github.com/argilla-io/argillaDistilabel: github.com/argilla-io/distilabelSynthetic Data Generator: github.com/argilla-io/synthetic-data-generatorSelfInstruct: arxiv.org/abs/2212.10560Magpie: arxiv.org/abs/2406.08464TextClassification: arxiv.org/abs/2401.00368RAG: arxiv.org/abs/2401.00368DEITA: arxiv.org/abs/2312.15685PersonaHub: arxiv.org/abs/2406.20094PersonaHub Exploration: youtube.com/watch?v=timmCn8Nr6gFine Personas: huggingface.co/datasets/argilla/FinePersonas-v0.1#how-it-was-builtFine-tune for function calling: huggingface.co/learn/agents-course/en/bonus-unit1/fine-tuningAgents Course: huggingface.co/learn/agents-course/unit0/introductionHuggingFace Reasoning Course: huggingface.co/reasoning-courseText2SQL on HuggingFace Datasets: youtube.com/watch?v=5LUZq7MHolA0:00 Welcome David and Ben!0:56 The Synthetic Data Generator3:25 Data Generation Algorithms11:32 PersonaHub15:15 Exploration in Data Generation22:05 Systems for Data Generation 27:20 DSPy and Distilabel33:30 Diversity in Data Generation39:10 Synthetic Image Generation44:50 Evolutionary Algorithms for Synthetic Data50:48 Prompt to Model59:25 Exciting Directions for AI
Text-to-SQL is dead: The next generation of querying is Agentic Weaviate • Vector Database 2025-03-12 | This paper (arxiv.org/abs/2502.00032) introduces a type of agentic querying called Function Calling that uses an LLM to structure queries using predefined function calls in JSON format, with optional arguments for search, filters, aggregation, and grouping. Along with that, it also tests out a bunch of different models with a new dataset, DBGorilla, designed to evaluate agentic querying techniques on real-world use cases.Weaviate also just released a Query Agent, designed based on some of the work in this paper, to handle advanced agentic querying out of the box, find out more here: weaviate.io/blog/query-agent▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Weaviate 1.29 Release Highlights & Outlook Weaviate • Vector Database 2025-03-10 | Join us for a technical walkthrough on some of the Weaviate 1.29 highlights. We dive into features like:• Multi-valued Vectors aka multi-vector embeddings based on colBERT, enabling you to increase the accuracy of your search results• A new B25 implementation based on BlockMax WAND, speeding up search performance,• General Availability of role-based access control (RBAC)• A new NVIDIA model integration• and much more!Read the full blog post to learn more: weaviate.io/blog/weaviate-1-29-release00:00 Introduction00:50 Agenda01:07 Multi-vector embeddings15:50 RBAC33:06 Acync Replication36:14 NVIDIA model integration38:35 BlockMax WAND41:40 2025 Weaviate Roadmap & Outlook▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Letta AI with Sarah Wooders - Weaviate Podcast #117! Weaviate • Vector Database 2025-03-03 | Hey everyone! Thank you so much for watching the 117th episode of the Weaviate podcast! In this episode, we dive deep into the cutting edge of AI agent development with Sarah Wooders, co-founder and CTO of Letta AI. Emerging from Berkeley's Sky Computing Lab, Sarah and her team have pioneered a revolutionary approach to stateful agents - AI systems that genuinely remember both you and themselves across extended conversations. The conversation explores how the groundbreaking MemGPT project evolved into Letta's comprehensive Agent Development Environment (ADE), which empowers developers to build truly persistent AI experiences. Sarah shares powerful insights on context management, memory prioritization, and the critical role of databases in agent architecture. Whether you're building AI systems or simply curious about where conversational AI is heading, this episode illuminates how the future of agents depends not just on their reasoning capabilities, but on their ability to maintain coherent identity and memory over time.Links:Get started with Letta Desktop: docs.letta.com/quickstart/desktopStateful Agents: letta.com/blog/stateful-agentsMemGPT: arxiv.org/abs/2310.08560MemGPT Explained! - youtube.com/watch?v=nQmZmFERmrgLLMs Can Easily Learn to Reason from Demonstrations Structure, not content, is what matters! - arxiv.org/pdf/2502.073740:00 Welcome Sarah!4:03 Stateful Agents8:28 Context Window Optimization9:13 Self-Optimizing Agents12:43 Agent Personas14:48 Gradient Descent and Prompts19:58 Tools and Function Calling21:53 Agent Development Environment (ADE)31:54 Databases for Agents36:28 LLM-Tool Use Examples37:03 Multi-Agent Systems41:28 Model Routers43:26 Reasoning Models and Context Windows47:53 Inspiration for ADE50:52 Bring your own Tools57:00 Future Directions for AI
Agent Experience with Matt Biilmann, Sebastian Witalec, and Charles Pierse - Weaviate Podcast #116! Weaviate • Vector Database 2025-02-27 | Hey everyone! Thank you so much for watching another episode of the Weaviate Podcast! I am SUPER excited to welcome Matt Biilmann, Co-Founder and CEO of Netlify, as well as Sebastian Witalec and Charles Pierse from Weaviate to discuss Agent Experience! You have probably heard about how you can connect LLMs to external software tools. This supercharges the capabilities of AI systems and what they can do. So what does that mean for you as a software developer?This podcast explores different ideas around designing software user experiences for Agents as well as Humans. How do we write documentation for Agents differently than Humans? How do we design REST or gRPC APIs, or programming languages clients, for Agents differently than Humans? llms.txt, JSON tool definitions, agents to agent, breaking changes, … there were so many interesting topics explored in this podcast! I really hope you find it useful! As always more than happy to discuss these ideas further with you!Read the blog here! - https://biilmann.blog/articles/introducing-ax/Chapters0:00 Weaviate Podcast #116!0:20 Welcome Matt, Sebastian, Charles1:45 Agent Experience10:20 Docs for Agents13:12 llms.txt17:20 JSON Tool Definitions21:16 Have your Agent talk to my Agent26:27 gRPC APIs29:56 Agentic Workflows33:04 LLM Bias in Software Vendors42:52 New APIs and Breaking Changes
How to pick the best LLM in 2025 Weaviate • Vector Database 2025-02-20 | Stop relying on OpenAI models for everything!Whether you need speed, quality, or performance, there's a perfect language model out there for your needs.But with so many choices, how do you pick the best one?Here are the 4 key factors to consider:• 𝗤𝘂𝗮𝗹𝗶𝘁𝘆: Look for an LLM that performs consistently across tasks like chatbot interactions, language understanding, and coding.Best options: DeepSeek R1, OpenAI's o1, OpenAI's o3 Mini• 𝗣𝗿𝗶𝗰𝗲: Consider the cost per token for both input and output.Best options: Gemini 2.0 Flash, GPT-4o Mini, and Mistral Small 3• 𝗦𝗽𝗲𝗲𝗱/𝗧𝗵𝗿𝗼𝘂𝗴𝗵𝗽𝘂𝘁: Generally, smaller models like Mistral Small 3 and GPT-4o Mini are faster than larger, high-quality models.Best options: Gemini 2.0 Flash, and GPT-4o Mini• 𝗢𝗽𝗲𝗻 𝗦𝗼𝘂𝗿𝗰𝗲: Models that allow for private useBest options: Llama 3, Microsoft Phi-4, Deepseek's latest models• 𝗖𝗼𝗱𝗶𝗻𝗴: Models that excel at code generationBest options: Claude 3.5 Sonnet, OpenAI's o3 Mini, Deepseek R1 & V3In this video, I used the Quality vs. Throughput and Price Diagram from artificialanalysis.ai/modelsTry it in @weaviate.io: weaviate.io/developers/weaviate/search/generative
Optimizing Retrieval Agents with Shirley Wu - Weaviate Podcast #115! Weaviate • Vector Database 2025-02-19 | Hey everyone! Thank you so much for watching the 115th episode of the Weaviate Podcast featuring Shirley Wu from Stanford University! We explore the innovative Avatar Optimizer—a novel framework that leverages contrastive reasoning to refine LLM agent prompts for optimal tool usage. Shirley explains how this self-improving system evolves through iterative feedback by contrasting positive and negative examples, enabling agents to handle complex tasks more effectively.We also dive into the STaRK Benchmark, a comprehensive testbed designed to evaluate retrieval systems on semi-structured knowledge bases. The discussion highlights the challenges of unifying textual and relational retrieval, exploring concepts such as multi-vector embeddings, relational graphs, and dynamic data modeling. Learn how these approaches help overcome information loss, enhance precision, and enable scalable, context-aware retrieval in diverse domains—from product recommendations to precision medicine.Whether you’re interested in advanced prompt optimization, multi-agent system design, or the future of human-centered language models, this episode offers a wealth of insights and a forward-looking perspective on integrating sophisticated AI techniques into real-world applications.Links:AvaTaR: arxiv.org/abs/2406.11200STaRK: arxiv.org/abs/2404.13207Learn more about Shirley Wu: https://cs.stanford.edu/~shirwu/Chapters0:00 Welcome Shirely!0:30 What is the state of AI?2:18 Graph Data Models and GraphRAG8:22 Unifying Textual and Relational Retrieval10:50 Multi-Vector Retrieval15:55 Embeddings and Properties21:58 PrimeKG27:25 AvaTaR: LLM Tool Use Optimization41:40 Complex Retrieval46:28 Agent Memory and DSPy55:00 Future Directions for AI
Contextual AI with Amanpreet Singh - Weaviate Podcast #114! Weaviate • Vector Database 2025-02-12 | Hey everyone! Thank you so much for watching the 114th episode of the Weaviate Podcast featuring Amanpreet Singh, Co-Founder and CTO of Contextual AI! Contextual AI is at the forefront of production-grade RAG agents! I learned so much from this conversation! We began by discussing the vision of RAG 2.0, jointly optimizing generative and retrieval models! This then lead us to discuss Agentic RAG and how the RAG 2.0 roadmap is evolving with emerging perspectives on tool use. Amanpreet continues to further motivate the importance of continual learning of the model and the prompt / few-shot examples -- discussing the limits of prompt engineering. Personally I have to admit I think I have been a bit too bullish on only tuning instructions / examples, Amanpreet made an excellent case for updating the weights of the models as well -- citing issues such as parametric knowledge conflicts, and later on discussing how Mechanistic Interpretability is used to audit models and their updates in enterprise settings. We then discussed Contextual AI's LMUnit for evaluating these systems. This then lead us into my favorite part of the podcast, a deep dive into RL algorithms for LLMs. I highly recommend checking out the links below to learn more about Contextual's innovations on APO and KTO! We then discuss the importance of domain specific data, Mechanistic Interpretability, return to another question on RAG 2.0, and conclude with Amanpreet's most exciting future directions for AI! I hope you enjoy the podcast!Learn more about Contextual AI! - contextual.aiContextual AI Platform: contextual.ai/blog/contextual-ai-platform-generally-availableLinks:Anchored Preference Optimization: arxiv.org/abs/2408.06266KTO: Model Alignment as Prospect Theoretic Optimization: arxiv.org/pdf/2402.01306RETRO: arxiv.org/abs/2112.04426Fusion-in-Decoder: arxiv.org/pdf/2007.01282RAG: arxiv.org/pdf/2005.11401LMUnit: contextual.ai/lmunitChapters0:00 Welcome Amanpreet!0:28 Founding Story and RAG 2.05:00 Agentic RAG10:25 The Limits of Prompt Engineering19:55 LMUnit Evals28:35 RL for LLMs40:00 Model Specialization46:12 Mechanistic Interpretability49:10 Back to RAG 2.055:15 Future Directions for AI
Think Deepseek-R1 was trained like other top LLMs? Think again. Weaviate • Vector Database 2025-02-07 | Try out Deepseek open source: github.com/weaviate/recipes/blob/main/weaviate-features/generative-search/generative_search_ollama/deepseek-ollama-epic-games-rag.ipynbPaper: arxiv.org/abs/2501.12948▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Cartesia AI with Karan Goel - Weaviate Podcast #113! Weaviate • Vector Database 2025-01-28 | Hey everyone! Thank you so much for watching the 113th episode of the Weaviate Podcast with Karan Goel from Cartesia AI! Cartesia AI is leading the AI world in text-to-speech models! As exciting as these new applications in speech generation are, Cartesia is also building around an incredibly exciting new neural network architecture that cuts across all of AI -- State Space Models. State Space Models (SSMs) present a new approach to modeling long sequences circumventing the quadratic attention bottlenecks of transformers. In the podcast, we discuss Karan's perspectives around end-to-end modeling, long context and Multimodal processing, building and deploying a new kind of model, and more! I hope you find the podcast interesting and useful! As always more than happy to discuss these ideas further with you! Thank you for listening!Cartesia AI: cartesia.aiIntroduction to State Space Models (SSMs): huggingface.co/blog/lbourdois/get-on-the-ssm-trainEfficiently modeling long sequences with structured state spaces by Albert Gu, Karan Goel, and Christopher Re: arxiv.org/pdf/2111.00396Karan Goel Google Scholar: scholar.google.com/citations?user=1i3X2GgAAAAJChapters:0:00 Welcome Karan!0:37 Founding Cartesia AI2:33 State Space Models (SSMs)9:04 Audio Data Deep Dive14:45 The launch of Sonic23:30 Inference vs. Agent APIs29:58 Learning, Search, and Reasoning34:35 RETRO and Fusion-in-Decoder RAG38:00 State in Sequence Models42:00 CUDA for SSMs49:00 Many Shot In-Context Learning50:35 AI-Native Text-to-Speech Applications
How to choose an embedding model Weaviate • Vector Database 2025-01-28 | How do you chose the best embedding model for your use case?(and how do they even work, anyways?)- Learn more in this upcoming webinar: https://hubs.la/Q031qM_h0- Weaviate Embeddings: weaviate.io/blog/introducing-weaviate-embeddings- MTEB Leaderboard: huggingface.co/spaces/mteb/leaderboard- How to chose an embedding model, on demand webinar: webinars.techstronglearning.com/simplify-building-ai-native-embedding-models-and-vector-databases- Word2Vec paper: arxiv.org/abs/1301.3781- Colorful vectors: huggingface.co/spaces/jphwang/colorful_vectors▬▬▬▬▬▬▬▬▬▬▬▬ CONNECT WITH US ▬▬▬▬▬▬▬▬▬▬▬▬- Visit http://weaviate.io- Star us on GitHub github.com/weaviate/weaviate- Stay updated and subscribe to our newsletter: newsletter.weaviate.io- Try out Weaviate Cloud for free here: https://console.weaviate.cloud/Got a question?- Forum: forum.weaviate.io- Slack: weaviate.io/slackConnect with us on- Twitter: twitter.com/weaviate_io- LinkedIn: linkedin.com/company/weaviate-io
Agentic RAG vs Vanilla RAG Weaviate • Vector Database 2025-01-20 | Vanilla RAG is a straightforward pipeline: retrieve, augment, and then generate.But, agentic RAG?It's the strategist — planning steps, reasoning, and fetching exactly what's needed, even from the web or multiple tools!Watch the full podcast: youtube.com/watch?v=Eh4uQq43jA4
