Generative AI can be incredibly powerful when it comes to legacy modernization. Not only can it help us better understand a large, aging codebase, it can even help us reverse engineer a legacy system when we don't have access to the complete source code. Doing it, though, requires a specific approach that's being described as 'context engineering'.

This is something we've been exploring a lot in recent months at Thoughtworks. On this episode of the Technology Podcast, Thoughtworks' lead for AI-enabled software engineering, Birgitta Böckeler, and tech principal Chandirasekar Thiagarajan join hosts Ken Mugrage and Neal Ford to discuss how it works.

They explain the process, the tools and what the work taught them about both generative AI and legacy modernization.

Read Birgitta's blog post on reverse engineering with AI: https://www.thoughtworks.com/insights/blog/generative-ai/blackbox-reverse-engineering-ai-rebuild-application-without-accessing-code

How should businesses go about actually navigating AI? It's one thing to strategize and generate new ideas, but what needs to be done to put it into practice in a way that's effective and commercially impactful?

In this episode of the Technology Podcast, new host Nigel Dalton is joined by his Thoughtworks colleague May Xu — Head of Tech for Thoughtworks APAC — and Simon Noonan, CTO at Australian business software company MYOB.

Thoughtworks has been working closely with MYOB for a number of years now; May and Simon explain how they collaborate and offer their perspectives on everything from leadership to architecture in a world where AI has become imperative.

Learn more about Thoughtworks' partnership with MYOB: https://www.thoughtworks.com/clients/myob

In an age of vibe coding and LLMs, do we really need to care about documentation? Do we need to spend time and energy producing it — time when we could just be shipping code? Of course we do; particularly if we want to communicate and share software with other humans.

To discuss documentation in 2025, Technology Podcast host Lilly Ryan is joined by Heidi Waterhouse, a very special guest with an esteemed and varied career in technical communcation.

In this episode, Lilly and Heidi tackle the challenges of documentation in a world increasingly infused with AI-generated code and text, explore whether prompt engineering is really just technical writing in disguise and examine the difficulties of writing for highly specific audiences.

They also cover Heidi's Progressive Delivery, an upcoming book about bridging the gap between software delivery and business value. It's due to be released in the latter part of 2025 and written alongside James Governor, Kim Harrison and Adam Zimman.

Find out more about Heidi Waterhouse by visiting her website: https://heidiwaterhouse.com/

Learn more about Progressive Delivery: https://itrevolution.com/product/progressive-delivery/

The technology industry has embraced specialisms — not just in different fields or job roles, like web development or security, but even in terms of particular platforms or stacks. But are we losing something as every tech professional is forced to push themselves into increasingly smaller niches?

Martin Fowler and Unmesh Joshi think so. They've been thinking a lot about the importance of what they call "Expert Generalists" — professionals who "can dissect unfamiliar challenges, spot first-principles patterns and make confident design decisions with the assurance of a specialist."

In this episode of the Technology Podcast, Martin and Unmesh join hosts Prem Chandrasekaran and Lilly Ryan to discuss how they came to identify the importance of expert generalists and why it was important to not just talk about the issue, but to explicitly name it. They also explore how they believe the industry can cultivate and encourage expert generalists, despite an entrenched tendency to overlook their value.

Read Martin and Unmesh's article, written with Gitanjali Venkatraman: https://martinfowler.com/articles/expert-generalist

Back in 1994, Peter Deutsch and his colleagues at Sun Microsystems identified what they described as the "eight fallacies of distributed computing" — flawed assumptions that often get made when teams move from monolithic to distributed software architectures. In recent years, software architecture experts and regular writing partners Neal Ford and Mark Richards have identified a further three new fallacies of distributed computing: versioning is easy; compensating updates always work; and observability is optional.

In this episode of the Technology Podcast, Neal and Mark join host Prem Chandrasekaran to talk through these three new fallacies, before digging deeper into other important issues in software architecture, including modular monoliths and governing architectural characteristics. Listen for a fresh perspective on software architecture and to explore key ideas shaping the discipline in 2025.

Learn more about the second edition of Neal and Mark's Fundamentals of Software Architecture: https://www.oreilly.com/library/view/fundamentals-of-software/9781098175504/

What if your AI agents could think more like IT operations staff — and less like tools?

In this episode, we catch up with Zichuan Xiong, to explore the Model Context Protocol (MCP) — a powerful new way to give AI agents deeper awareness of the tools, information and history they need to work effectively in the operations space. Unlike traditional APIs that just trigger functions, MCP adds a semantic layer of context that helps AI understand what to do, why it matters and how to do it better.

Whether you’re deep in site reliability engineering (SRE) or just curious about the next leap in AIOps, this episode unpacks how MCP could be the missing layer between today’s tools and tomorrow’s autonomous systems.

If you want to find out more, check out this piece by Zichuan at al, https://www.thoughtworks.com/insights/blog/machine-learning-and-ai/mcp-critical-ai-driven-sre

Google I/O 2025 took place in May. It's always a great opportunity to find out how Google is trying to shape the industry agenda, but this year the predominance of Gemini meant the event was a chance to get a better look at how Google will play its hand in the AI market in the months to come.

To dissect the headlines from this year's Google I/O and explore what we can learn about Google's strategic focus — and how the company is thinking about AI — host Ken Mugrage is joined by Andy Yates on the Technology Podcast. As Head of Ecosystems Development at Thoughtworks, Andy plays an important role in helping the organization and its clients undertstand, analyze and engage with the major platforms and vendors.

This edition of Google I/O, he explains, was significant and particularly useful for helping us understand how the world is going to be consuming AI products and services as the technology becomes more and more embedded in the mainstream.

Read more of Andy's perspective on Google I/O 2025 on the Thoughtworks blog: https://www.thoughtworks.com/insights/blog/technology-strategy/google-io-2025-key-takeaways

Mainframe modernization is hard: there's a huge amount of complexity that needs to be understood before it can be effectively addressed. Generative AI, however, can be a particularly powerful tool for understanding mainframe legacy codebases, something we've been exploring with Mechanical Orchard while working together on its Imogen modernization platform.

In this episode of the Technology Podcast, hosts Ken Mugrage and Alexey Boas are joined by Thoughtworks CTO Rachel Laycock and Mechanical Orchard CEO and Founder Rob Mee to discuss the partnership between the two organizations. They discuss how the collaboration began, the challenges of leveraging generative AI tools for such risky projects and what the wider implications are for AI in software engineering. Listen for a fresh perspective on both legacy modernization and generative AI.