In this episode, Chris Churilo and Jay Byoun discuss the intricacies of vector databases, particularly focusing on Zilliz and Milvus. They explore the performance and scale requirements for vector databases, the challenges customers face in production, and the importance of understanding vector counts. The conversation delves into user experience, query challenges, and the balance between performance, cost, and accuracy. They also touch on hybrid search, metadata filtering, and the future of vector databases in relation to data lakes and security considerations.

In this podcast episode, Chris Churilo (VP of Marketing) and James Luan (VP of Engineering at Zilliz) discuss the advancements and features of Milvus, a scalable vector database designed for unstructured data. They delve into the latest version, Milvus 2.6, highlighting its focus on cost reduction, scalability, and integration with AI applications. The conversation covers various topics, including tiered storage, quantization techniques, and the introduction of a new logging solution called Woodpecker. Additionally, they explore the importance of embedding models and user-defined functions, as well as optimizations in scalar search and phrase matching.

To learn more about working with unstructured data, visit ⁠⁠zilliz.com⁠⁠ and ⁠⁠milvus.io⁠⁠

In this conversation, Stefan Webb (Developer Advocate) and Jay Byoun (Solutions Architect) discuss the intricacies of Zilliz Cloud and vector databases, focusing on indexing algorithms, recall and latency trade-offs, chunking strategies, cost considerations in LLM inference, and diverse use cases for vector databases. They also provide insights on how to get started with Zilliz, emphasizing the importance of understanding these concepts for effective data management and application development.

To learn more about working with unstructured data, visit ⁠zilliz.com⁠ and ⁠milvus.io⁠

In this episode, Stefan Webb (developer advocate at Zilliz) interviews Cameron Pfiffer from .txt, discussing the intersection of generative AI, developer advocacy, and structured output. Cameron shares his journey into the field, the mission of .txt, and the importance of constrained sampling in AI development. The conversation explores the functionalities of Outlines, a tool designed to enhance structured output, and its future developments. Additionally, they discuss the integration of vector databases with constrained output and how listeners can engage with the community and resources available.

To learn more about working with unstructured data, visit zilliz.com and ⁠milvus.io.

Stephen Batifol and Stefan Webb, Developer Advocates at Zilliz (the maintainers of the open source vector database Milvus), explore the increasingly important world of GenAI model evaluation. They discuss why it is important to evaluate your RAG or Agent pipeline during design and after deployment, how to do it, and where things are headed. Learn about the experimental mindset, LLM-as-a-Judge, RAGAS library, and other exciting topics!

To learn more about working with unstructured data, visit ⁠zilliz.com⁠ and ⁠milvus.io⁠.

Stefan Webb and Stephen Batifol, Developer Advocates at Zilliz (the maintainers of the open source vector database Milvus), dive deep into the evolving world of AI agents—what they are, how they work, and their growing impact on technology and society.

They start by breaking down how agents operate and their workflows, then explore real-world applications and the challenges they face across industries. Beyond the technical, they discuss the philosophical and ethical implications of automation, questioning what it means for human work, creativity, and decision-making.

For builders, Stefan and Stephen cover the essential tools and techniques for developing AI agents, along with how to evaluate their performance and where the future of autonomous systems is headed. Whether you're an engineer, a strategist, or simply curious about AI, this series will give you the insights you need to navigate the agent-driven future.

To learn more about working with unstructured data, visit zilliz.com and milvus.io.