Unlearning, experimentation and engineering rigor in an agentic world

Managing distributed systems and complex workflows can be challenging. What happens when something fails? If a task isn't executed to completion, that can lead to serious problems. From transaction and billing failures to deploying software, even small issues can have significant consequences. This is one of the reasons for durable computing. Designed to isolate code from crashes, it preserves state so a task is completed even when something fails. To discuss durable computing, explore why it matters today and how we've been using it at Thoughtworks, Brandon Cook and John Coleman join host Alexey Boas on the Technology Podcast. They dive into the current platform ecosystem and what it means for developers — and requires of them. They also talk about the value of durable computing for AI, explaining why the concept of 'durable agents' offers an important of avenue of investigation in a world eager to embrace agentic systems.  Learn more about durable computing in this blog post from July 2025: https://www.thoughtworks.com/insights/blog/cloud/durable-computing-making-easier-resilience-distributed-systems  

In January 2026, Thoughtworks launched AI/works™, an agentic development platform. It promises to make the capabilities of AI agents a reality for the enterprise, helping in areas including understanding complex legacy code, forward engineering new software solutions and agent governance. How, though, does it actually work in practice? And what does it mean for the organizations and teams Thoughtworks works with? In this episode of the Technology Podcast, new host Rickey Zachary is joined by Bharani Subramaniam (CTO for Thoughtworks India and the Middle East) and Shodhan Sheth (Head of Enterprise Modernization, Platforms and Cloud) to discuss AI/works™, taking in how the platform emerged from a number of recent Thoughtworks projects to how it's delivering value to businesses today.  As well as an inside perspective on Thoughtworks' new platform, the episode also offers a deep and timely exploration of questions and challenges the rapid rise of AI agents in software engineering has surfaced across every part across industry. Learn more about AI/works™: https://www.thoughtworks.com/ai/works

In a world that's being transformed by AI agents and agentic systems, how do software developers unlearn what they know while also maintaining engineering rigor? In an in-person conversation with Nathen Harvey, Developer Relations Engineer at Google Cloud, and Patrick Debois, Developer Relations at Tessl, host Ken Mugrage dives into the ways individuals, teams and organizations are walking the line between experimentation and well-established engineering practices as they seek to innovate while ensuring resilience, reliability and security. Thoughtworks is a platinum sponsor of the 2025 DORA report: https://www.thoughtworks.com/en-us/insights/reports/the-2025-dora-report  

If AI agents really are the future of how work will be done — in software engineering and beyond — the platforms on which they are built, run and maintained will be crucial. This is a topic two Thoughtworkers, Ben O'Mahony and Fabian Nonnenmacher, are currently writing about. Although not due to be published until early 2027, the first two chapters of Building AI Agent Platforms are now available as part of O'Reilly's Early Release scheme. Its goal is to provide readers with a complete roadmap for developing AI agent platforms, from agent development to architectural principles to observability and governance.  In this episode of the Technology Podcast, the authors speak to regular host Ken Mugrage about the book, why agent platforms are a critical part of any AI strategy and some of the challenges of developing and maintaining them. Listen for an early look at what looks set to be a valuable book in the world of AI development and to gain a clearer perspective on what agentic AI really means at the start of 2026. Learn more about Building AI Agent Platforms: https://www.oreilly.com/library/view/building-ai-agent/0642572243906/  

You can grasp the basics of software architecture by learning design patterns, but you probably won't master it — to do that you have to get to grips with antipatterns too. Often these lessons are hard-won through experience, derived from seeing what happens when architectural decisions (or the lack of them) collide with the messy reality of the real world.  While there's obviously no replacement for practical experience, Neal Ford, Mark Richards and Raju Gandhi want to share theirs to give architects the opportunity to avoid common antipatterns and pitfalls and make smarter decisions. That's what they're doing with their forthcoming book (due September 2026) Architecture Patterns, Antipatterns and Pitfalls. On this episode of the Technology Podcast, host Neal is joined by his co-authors and their editor from O'Reilly, Sarah Grey, to discuss the book and to explore why getting to know on antipatterns and pitfalls matters as much as learning design patterns. Learn more about Software Architecture Patterns, Antipatterns and Pitfalls: https://www.oreilly.com/library/view/software-architecture-patterns/0642572221119/ Listen to Neal and Mark discuss the various intersections of software architecture on an episode of the Technology Podcast from 2025: https://www.thoughtworks.com/insights/podcasts/technology-podcasts/exploring-intersections-software-architecture

The 'age of intent' is a phrase that's been around for a number of years. However, with the rise of AI agents in 2025 it has the potential to become a key trend for 2026. It describes a new way of thinking about digital interaction in which the gap between human intention and output are reduced even further through AI assistance. Thoughtworks' APAC CTO Sarah Taraporewalla has been exploring the age of intent in recent months; she's written a series of blog posts that tackle what this new phase of digital interaction means for businesses and how they can prepare themselves. On the latest episode of the Technology Podcast, Sarah joins host Lilly Ryan to discuss the concept the age of intent and its implications for the future of digital experiences.  Read Sara Taraporewalla's series on the age of intent: The interface is dead. Time for the age of intent: https://www.thoughtworks.com/insights/blog/generative-ai/the-interface-is-dead-time-for-the-age-of-intent From prototype to transformation: https://www.thoughtworks.com/insights/blog/generative-ai/the-age-of-intent-from-prototype-to-transformation What it takes to become an intent-ready organization: https://www.thoughtworks.com/insights/blog/generative-ai/the-age-of-intent-from-prototype-to-transformation Rethinking value in the AI economy: https://www.thoughtworks.com/insights/blog/generative-ai/the-age-of-intent-rethinking-value-in-AI-economy        

This year's DORA report focuses on AI-assisted software development. While one of the key themes is just how ubiquitous AI is today in software engineering, that's only part of the picture. In fact, the report outlines many of the challenges the adoption of these technologies are posing and explores the barriers and obstacles that need to be addressed to ensure AI-assistance leads to long-term success. In this episode of the Technology Podcast, host Ken Mugrage is joined by Chris Westerhold — Global Practice Director for Engineering Excellence at Thoughtworks — to discuss this year's DORA report (for which Thoughtworks is a Platinum sponsor). They dive into some of the reports findings, and explore the risks of increasing throughput, the changing demands on software developers, the importance of developer experience and how organizations can go about successfully measuring AI impact. You can find the 2025 DORA report here: https://cloud.google.com/resources/content/2025-dora-ai-assisted-software-development-report Read Chris Westerhold's article on this year's findings: https://www.thoughtworks.com/insights/articles/the-dora-report-2025--a-thoughtworks-perspective  

Vibe coding was, remarkably, named word of the year by the Collins English Dictionary at the start of November 2025 — pretty good going for a term that was only coined in February. We first discussed it on the Technology Podcast back in April, and, given its prominence in the collective lexicon this year, thought we should revisit and reflect on the topic as 2025 draws to a close.  Lots has happened in the intervening months: MCP adoption, the evolution of agentic coding tools and practices like context engineering have had a significant impact on the way the world is thinking about and using AI.  To talk about it all and reflect on the implications, Thoughtworkers and regular podcast hosts Prem Chandrasekaran, Lilly Ryan and Neal Ford reconvened for a follow up to our April conversation. Taking in everything from the term's semantic slipperiness, its security risks and the challenges of maintaining AI-generated code, this is a discussion that, despite going deep into vibe coding, also touches on a huge range of issues in the technology industry today. Before we enter 2026, looking back on the good, the bad and the ugly of the last 12 months of experimentation is essential if we're to build better software for the world in the future. This episode aims to be a guide through that process. Listen to our April episode on vibe coding: https://www.thoughtworks.com/insights/podcasts/technology-podcasts/vibe-coding Read Ken Mugrage's blog post exploring the shift from vibe coding to context engineering in 2025: https://www.thoughtworks.com/insights/blog/machine-learning-and-ai/vibe-coding-context-engineering-2025-software-development

One of the biggest stories in software engineering in 2025 is the impact of generative AI on the software development lifecycle. From advances in coding assistance to the emergence of so-called agentic coding, there's undoubtedly a lot for software developers to process, learn and experiment with — not to mention rapid change to contend with. On this episode of the Technology Podcast, host Ken Mugrage is joined by Brandon Cook to discuss not only how AI has been shaping the way software developers work but how developers can play an active role in ensuring the technology is leveraged safely and successfully. Taking in everything from sensible defaults and best practices to evaluating how much autonomy you should give up to an agent in any given problem, this episode offers both a snapshot of where we are today and the role we all have to play in deciding what the future will look like. Explore the Thoughtworks Technology Radar: thoughtworks.com/radar Listen to Brandon's last appearance on the Technology Podcast from July 2024: https://www.thoughtworks.com/insights/podcasts/technology-podcasts/sensible-defaults-way-think-technology-practices  

In every Thoughtworks Technology Radar we feature three to five themes that represent the core issues and topics that emerged from the conversations we had when putting the publication together. This time (Fall 2025) they're all united by AI. They are: infrastructure automation arriving for AI, the rise of agents elevated by MCP, AI coding workflows and emerging AI antipatterns. On this episode of the Technology Podcast, Bryan Oliver joins Neal Ford and Ken Mugrage to discuss all four of volume 33's themes. They dive into what they mean, how the team arrived at them and what they tell us about the state of software engineering and AI in 2025. Read the latest Thoughtworks Technology Radar: thoughtworks.com/radar Volume 33 will be published November 5, 2025.