The Stack Overflow Podcast

Connecting the dots for accurate AI

31 min · 12. maj 202631 min
episode Connecting the dots for accurate AI cover

Beskrivelse

At HumanX, Ryan is joined by Philip Rathle, CTO at Neo4j to discuss what knowledge context means for AI agents, how limitations like stale training data make the model-only approach to agents a bad fit for enterprise environments, and how Graph RAG raises the bar for accuracy and reduces context rot by combining vectors with a knowledge graph so agents are more targeted and connected. Episode notes: Neo4j [https://neo4j.com/] is a native graph database management system designed to handle complex, highly-connected data by focusing on relationships rather than tables. You can try it out for free on Aura [https://neo4j.com/product/auradb/] and learn more at their Graph Academy [https://graphacademy.neo4j.com/categories/workshops/].  Connect with Philip on LinkedIn [https://www.linkedin.com/in/prathle/].  See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

Kommentarer

0

Vær den første til at kommentere

Tilmeld dig nu og bliv en del af The Stack Overflow Podcast-fællesskabet!

Kom i gang

2 måneder kun 19 kr.

Derefter 99 kr. / måned · Opsig når som helst.

  • Podcasts kun på Podimo
  • 20 lydbogstimer pr. måned
  • Gratis podcasts
Kom i gang

Alle episoder

942 episoder

episode Connecting the dots for accurate AI cover

Connecting the dots for accurate AI

At HumanX, Ryan is joined by Philip Rathle, CTO at Neo4j to discuss what knowledge context means for AI agents, how limitations like stale training data make the model-only approach to agents a bad fit for enterprise environments, and how Graph RAG raises the bar for accuracy and reduces context rot by combining vectors with a knowledge graph so agents are more targeted and connected. Episode notes: Neo4j [https://neo4j.com/] is a native graph database management system designed to handle complex, highly-connected data by focusing on relationships rather than tables. You can try it out for free on Aura [https://neo4j.com/product/auradb/] and learn more at their Graph Academy [https://graphacademy.neo4j.com/categories/workshops/].  Connect with Philip on LinkedIn [https://www.linkedin.com/in/prathle/].  See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

12. maj 202631 min
episode AI giveth and AI taketh CPU cover

AI giveth and AI taketh CPU

Recorded on the floor of HumanX, Ryan is joined by AMD CTO Mark Papermaster to discuss AMD’s silicon strategy for AI borne of their long history of heterogeneous CPU/GPU computing, how chipmakers are dealing the wide range of AI workloads from training to inference, and the paradox of agents both eating up all the compute and helping AMD accelerate chip innovation.  Episode notes:  Want to learn more about the topics Mark and Ryan discussed in this episode? Check out the AMD Advanced Insights podcast [https://open.spotify.com/show/37Pa02uPo0aN25I8PzgdxW], a monthly show hosted by Mark. Connect with Mark on LinkedIn [https://www.linkedin.com/in/mark-papermaster-66914925/].   TRANSCRIPT [https://stackoverflow.blog/2026/05/08/ai-giveth-and-ai-taketh-cpu/] See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

8. maj 202632 min
episode What (un)exactly do you mean by semantic search? cover

What (un)exactly do you mean by semantic search?

Ryan welcomes Bryan O’Grady, Head of Field Research and Solutions Architecture at Qdrant, to discuss the differences between traditional text search engines powered by Lucene and modern vector databases, when vector search’s exact-match needs work for things like logs and security analytics and when semantic search works for user-facing discovery and non-exact results, and how Qdrant is growing into video embeddings and local-agent contexts.  Episode notes:  Qdrant [https://qdrant.tech/] offers high-performance vector search at scale with any deployment model. Connect with Brian on LinkedIn [https://www.linkedin.com/in/brian-ogrady/] or email the Qdrant team at support@qdrant.io [support@qdrant.io].  Congratulations to user Brad Larson [https://stackoverflow.com/users/19679/brad-larson] for winning a Populist badge for their answer to Find the tangent of a point on a cubic bezier curve [https://stackoverflow.com/questions/4089443/find-the-tangent-of-a-point-on-a-cubic-bezier-curve]. TRANSCRIPT [https://stackoverflow.blog/2026/05/05/what-un-exactly-do-you-mean-by-semantic-search/] See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

5. maj 202628 min
episode Time is a construct but it can still break your software cover

Time is a construct but it can still break your software

Ryan welcomes Jason Williams, senior software engineer at Bloomberg and the creator of Rust-based JavaScript engine Boa, to the show to dive into why date and time handling in JavaScript is so difficult and how the Temporal proposal aims to fix it. They explore the current flaws and issues in JavaScript that make the Date object so hard to work with, how libraries like Moment.js helped but eventually became too complex themselves, and why the Temporal proposal took nine years to complete.  Episode notes:  Temporal [https://tc39.es/proposal-temporal/docs/] is a new TC39 proposed standard for JavaScript that replaces the Date object. It operates as a top-level namespace and brings a modern date/time API to the ECMAScript language. Connect with Jason on Bluesky [https://bsky.app/profile/jason-williams.co.uk] or at his website [https://jason-williams.co.uk/].  Congrats to Great Answer badge winner BrenBarn [https://stackoverflow.com/users/1427416/brenbarn], who won the badge for their answer to rethrowing python exception. Which to catch? [https://stackoverflow.com/questions/25001971/rethrowing-python-exception-which-to-catch]. TRANSCRIPT [https://stackoverflow.blog/2026/05/01/time-is-a-construct-but-it-can-break-your-software/] See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

1. maj 202635 min
episode Your LLM issues are really data issues cover

Your LLM issues are really data issues

Ryan welcomes Harsha Chintalapani, co-founder and CTO at Collate and co-creator of Open Metadata, to the show to discuss why AI and LLMs struggle with real-time, structured production data. They explore how schema changes, inconsistent definitions (like “customer”), and weak governance can break both your analytics and MLs, and what companies can do to get their data AI-ready, from metadata management to observability.  Episode Notes:  Collate [https://www.getcollate.io/] is a semantic intelligence platform built on a semantic metadata graph for discovery, governance, and AI observability across your data ecosystem. Connect with Harsha on LinkedIn [https://www.linkedin.com/in/sriharsha/].  Congrats to user buttonsrtoys [https://stackoverflow.com/users/2079612/buttonsrtoys], who won a Famous Question badge for their question Possible to edit PDF without embedded font installed? [https://stackoverflow.com/questions/27807875/possible-to-edit-pdf-without-embedded-font-installed]. TRANSCRIPT [https://stackoverflow.blog/2026/04/28/your-llm-issues-are-really-data-issues/] See Privacy Policy at https://art19.com/privacy [https://art19.com/privacy] and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info [https://art19.com/privacy#do-not-sell-my-info].

28. apr. 202631 min