Season 3: Episode 12 - PureGym's Engineering Director on AI, Scale & Why Every Engineer Is a Manager
The CTO LensShow Notes
The CTO Lens Podcast is back with Episode 12 of Season 3, and this week David and Sophie are deep into another packed agenda - rogue AI agents, disaster recovery, recruitment bias and a Wheel of Wonder that nobody saw coming.
For our main segment in this episode, Sophie and David sit down with Gordon Lucas, Group Software Engineering Director at PureGym - the UK's largest gym chain with over 600 sites and now expanding into the US - for a conversation about what it really means to lead an engineering team through rapid business transformation. Gordon shares how he has been quietly mandating AI adoption across his team for two years, why he believes every engineer is now effectively a manager, and what happens to culture and craft when the code starts writing itself.
For the news section, David and Sophie cover the growing pattern of AI agents escaping their sandboxes - this time it's Meta's turn, as their cybersecurity agent breaks out during testing and the blame game begins. Anthropic's Mythos is revealed to have created fake accounts based on real people, sent malicious files through GitHub and then edited its own activity history to cover its tracks - stopped only by human review. There is also a 90-second power outage in Manchester that brought the rail network to its knees - and David asks whether any of us are actually testing our disaster recovery plans. Elon Musk tells SpaceX's first ever earnings call that Starlink will deliver most of the world's internet within 10 years, while his executives quietly walk it back. And a BBC investigation finds that AI recruitment tools may be systematically screening out women returning to work - and that some are already facing legal action for it.
And in the Flip Side, the Wheel of Wonder makes its debut - spinning to land on the question of whether engineers should be allowed to choose their own tech stack. The coin decides who argues what. David ends up convincing himself. They call it a tie.
More from The CTO Lens
In this week's episode of The CTO Lens, David is back from holiday and Sophie is fresh from the London Tech Leaders Summit, as they tackle a familiar challenge for technology leaders: dealing with non-technical executives and boards. With technology shaping almost every business decision, is it still acceptable for senior leaders to say they don't understand it? Or is the bigger risk an executive who thinks owning an iPhone makes them a technologist? David and Sophie explore where responsibility lies, how to bridge the gap between technology and the boardroom, and why a little curiosity goes a long way. (Expect an analogy involving jam doughnuts!) In the news, Starling Bank's move towards sovereign AI raises questions about whether building your own AI model is realistic for organisations without billions of transactions at their disposal. Meanwhile, Anthropic's latest workforce report reveals a shift in how employees work, with fewer junior hires, more meetings and some staff feeling more like AI supervisors than creators. And with growing scrutiny of social media platforms, should technology leaders be paying closer attention to the wider societal impact of AI? Plus, David and Sophie explore TIME's Best Inventions of 2026, from a tactile tablet helping blind sports fans follow the action to a robotic beekeeper that David, an amateur beekeeper himself, isn't entirely convinced by.
Sophie continues season three live from the London Tech Leaders Leadership Summit, bringing five more back-to-back conversations with tech leaders on stories and lessons from across their careers - from rethinking the technologist skill set to accessibility as a business imperative to how boards should prioritise change. This is part two. How AI Is Changing the Technologist Skill Set
Sophie closes out season three live from the London Tech Leaders Summit 2026. Part one features four back-to-back conversations with tech leaders on stories and lessons from across their careers. Part two follows on Thursday. From leading agentic development for 3,500 developers in a large corporation to building hands-on again at a startup
Sophie and David are back for episode 19, and once again there's no guest this week - which means it's straight into the news, followed by two of their own segments. David and Sophie spend time on two segments of their own for this episode given no guest this week. First, they dig into the idea of the "accidental CIO" - the notion that many of today's technology leaders got there through years of varied, messy, hands-on experience rather than a deliberate career plan, and what might be lost as the next generation is shaped instead by certifications and narrow specialisms. Then, in a rapid-fire closing segment, they're each handed ten pairs of essential leadership qualities and forced to keep only one from each pair - ambition or empathy, vision or execution, your biggest customer or your biggest engineer - arguing their way to a final top ten qualities for the tech leader of the future. In the news: OpenAI's AI agents were behind what cybersecurity professionals are calling the first reported AI agent-led government website hack anywhere in the world, after breaching both the Australian and US government systems. New research suggesting AI-driven efficiencies in US healthcare insurance may actually be driving costs up rather than down, and what that means for tech leaders trying to understand the wider systems their technology sits inside. A cybersecurity firm's discovery that around 16,000 Supabase-hosted databases - many of them vibe-coded - have been exposing personal data to the open web. And, for something lighter, Google's new AI-powered virtual wardrobe app that finally makes Cher's outfit computer from Clueless a reality.
Sophie and David are back for episode 18. This week, they've set themselves a challenge: run through the week's news without mentioning AI once (it does not entirely go to plan) and dispel five leadership myths. The five leadership myths cover whether digital transformation really belongs to IT, whether titles actually command influence, whether lines of code equal productivity, whether remote teams need micromanaging, and whether a great product can sell itself - David and Sophie agree on some, and push back hard on others. Among the news stories they did find: a BBC piece on social media addiction in over-55s and what it means for tech leaders to build KPIs that account for genuine harm, not just engagement. AWS's quiet admission - six months after airstrikes damaged its Bahrain and UAE infrastructure during the Iran war - that it has permanently lost some customer data, and what that means for how technology leaders actually calculate acceptable risk. The Revolut breach, where nearly 700 customers' sensitive data was handed over to hackers posing as Italian law enforcement, and the case for building a "sleep on it" delay into how teams respond to urgent-sounding requests. And Salesforce's Dreamforce conference outage, which took down the platform for over seven hours while staff were on stage praising its reliability.