JUXT Cast

S7E3 – Episode Notes There's a rush to build harnesses and scaffolding for AI agents, as though it were a problem no one had faced before. But we've spent decades learning how to build platforms for people. Isn't an AI agent just a demanding new kind of user? In this JUXTCast episode, JUXT CTO Henry Garner speaks with seasoned platform engineering leader Rachael Wonnacott about cutting through the AI noise with data and what her years of platform engineering experience tell us about building for AI agents. Along the way: why context is the new code, how a golden path can become a golden cage, and what a medieval eye remedy reveals about the reach of AI. Happy listening! Support JUXT Cast by contributing to their tip jar: https://tips.pinecast.com/jar/juxt-cast Send us your feedback online: https://pinecast.com/feedback/juxt-cast/bef2328d-a8ff-448a-9015-fd38206659c6 This podcast is powered by Pinecast.

The frontier of AI is focused on the next model and the next capability. But the AI we already have is sufficient to sift through the decades of rules and regulations that have grown too tangled for any one person to hold in their head. In this JUXTCast episode, JUXT CTO Henry Garner speaks with Mike Hsu, former US banking regulator and Acting Comptroller of the Currency, about what happens when code and policies drift from the context that gave them meaning. Both are written in good faith, for a particular moment, but the world moves on. What often remains is something we can't be sure still serves us, and don't dare to change. In software, we call this legacy code. Mike calls it policy sludge. And the remedies, it turns out, have a lot in common. Along the way: how law and software are heading in opposite directions, what overlapping rules have to do with a sixty-year-old programming language, and why cleaning all of this up comes down to trust. Happy listening! Send us your feedback online: https://pinecast.com/feedback/juxt-cast/2d695de3-4221-45b5-84b8-192292d9bb69

Episode Notes Most of the excitement around coding agents is about writing new code. But what about the code we already have? In this JUXTCast episode, JUXT CTO Henry Garner speaks with Chris Ford (ThoughtWorks), author of the O'Reilly book Agentic Engineering at Scale, about pointing coding agents at large legacy systems. Along the way: what after-dinner mints have to do with modernising them, why a good spec is never just markdown and whether AI will really make our software any better. Check out Chris Ford's book: https://learning.oreilly.com/library/view/agentic-engineering-at/0642572344306/ Chris Fird's Linkedin: https://www.linkedin.com/in/ctford/ Happy listening! Support JUXT Cast by contributing to their tip jar: https://tips.pinecast.com/jar/juxt-cast Send us your feedback online: https://pinecast.com/feedback/juxt-cast/b969809c-e85a-47b6-9b45-0efc631ad771

Episode Notes Open Source Readiness: https://osr.finos.org/docs/bok/Introduction http://osr.finos.org/FINOS AI Governance Framework: https://air-governance-framework.finos.org/ https://air-governance-framework.finos.org/ https://air-governance-framework.finos.org/Common Cloud Controls: https://ccc.finos.org/http://ccc.finos.org/ Get an early-access copy of Risk-First Software Development: https://riskfirst.org/Risk-First-Second-Editionhttps://riskfirst.org/Risk-First-Second-Edition

Episode Notes David Kelly is the author of several books, including Handbook of Model Risk Management for Vendors: https://www.linkedin.com/posts/davidkerrkelly_new-publication-handbook-of-model-risk-management-activity-7351188030193295360-nTur/. JUXT CTO Henry Garner will be presenting "Accelerating FINOS AI Governance Framework Adoption in Financial Services" at the upcoming FINOS conference on October 21st-22nd, 2025, in New York.

Episode Notes In this episode of the JUXTCast, we are joined by Malcolm Sparks (JUXT CTO), Henry Garner (Principal Engineer), Denis Lobanov (Senior Software Engineer), and Joe Littlejohn (Chief Delivery Officer) for an inside look at JUXT’s upcoming AI Radar. Initiated by Henry, the AI Radar is a new initiative designed to organise and share JUXT’s insights, advice, and recommendations on AI-related frameworks, tools, languages, and platforms. The team discusses the goals behind the project, key takeaways from their research so far, and how this evolving radar will support developers and decision-makers navigating the fast-moving AI landscape. Subscribe to our channel and stay tuned as the AI Radar will be available soon! Happy listening!

Episode Notes From settling into JUXT’s brand-new office to sharing insights from the XT25 Fintech Conference in London, the team discusses how the industry is evolving and how AI is shaping the future of software, finance, and developer tooling. Whether you're curious about what’s next for JUXT or the broader fintech and AI landscape, this episode sets the tone for an exciting new chapter.

Episode Notes In this engaging episode of the JUXT Cast, Jeremy Taylor and Malcolm Sparks sit down with Ryan Robitaille, the founder of Rabbit, https://github.com/ryrobes/rvbbit. Ryan shares his unique journey—from working with Oracle systems as a young Solutions Engineer to becoming a creative force in the world of data visualization. Ryan explores his experience building Rabbit, a tool designed to bridge the gap between proprietary BI tools and custom-built engineering solutions. Frustrated by the limitations of traditional tools like Tableau, Ryan envisioned a platform that offers the "best of both worlds": the simplicity of drag-and-drop dashboards with the power and flexibility of live coding and version control. Key Takeaways from the Episode: The Origins of Rabbit: How Ryan’s passion for combining artistry and data engineering sparked the creation of a platform that feels like a "game engine for data." Balancing Build vs. Buy: Insights into the perpetual organizational dilemma of purchasing BI tools versus building in-house solutions. Clojure’s Role: How Clojure and its philosophy of "code is data" played a pivotal role in Rabbit's architecture and flexibility. The Tableau Experience: Ryan reflects on Tableau's transformative early days and where the tool has hit its limits. Empowering End-Users: Why Ryan believes tools should offer a low bar for entry but a high ceiling for complexity. With fascinating anecdotes and deep technical insights, this episode sheds light on how data platforms can evolve to empower creativity, transparency, and collaboration.

Episode Notes This latest episode of the JUXTCast features Gene Kim, a Wall Street Journal bestselling author, celebrated researcher, and multiple award-winning Chief Technology Officer. Gene is widely recognized for his contributions to the DevOps movement and for co-authoring influential works such as The Phoenix Project and The DevOps Handbook. In this engaging discussion, Gene reflects on his career journey, from his time as the founder and CTO of Tripwire to his rediscovery of the joy of programming through Clojure. The episode explores key themes including high-performing technology organizations, the transformative role of AI in programming, and the strategic importance of modularity in systems design. The conversation also offers unique insights into the evolving role of AI in augmenting developer productivity and creativity. Gene shares his hands-on experience with pair programming and discusses the intersection of REPL-based programming, economic principles in software design, and the future of junior developers in an AI-enhanced ecosystem. Thoughts on a “DORA for GenAI and developers” study: https://x.com/RealGeneKim/status/1856146004724330862  2 hour pair programming with Steve Yegge! https://twitter.com/RealGeneKim/status/1860507119096869363 Description of what I did while walking dog: https://twitter.com/RealGeneKim/status/1853860996689064211  “From Naptime to Big Sleep: Using Large Language Models To Catch Vulnerabilities In Real-World Code,” https://googleprojectzero.blogspot.com/2024/10/from-naptime-to-big-sleep.html?m=1 XTDB: https://docs.xtdb.com/quickstart/sql-overview.html

Episode Notes In this episode, JUXT’s CEO Jon Pither, CTO Malcolm Sparks, and Head of Delivery Joe Littlejohn, are joined by guest Software Engineer Jake Howard to engage in a thoughtful discussion on the enduring static vs dynamic typing debate. While static typing has long been a staple in programming, the conversation leans toward the growing appeal of dynamic typing in modern software practices. The team explores how dynamic typing allows for quicker iteration, greater flexibility, and better adaptation to shifting project demands. They also take time to weigh the structure and reliability that static typing provides, making for a balanced look at both approaches.