At 9am, in the midst of London heatwave, one of the beautiful rooms at the Royal Society of Chemistry filled up with public sector digital leaders.
When we designed this event, I leant on my former career as a digital leader in the public sector. I asked myself one question: what would I be wrestling with right now if I were back as CIO at Parliament?
Two hours later, I had more answers than I'd bargained for.
99% of us don't know what we're doing.
Jamie Bartlett, author of the Sunday Times bestseller How to Talk to AI (and How Not To), opened our panel with a number that immediately grabbed the audience's attention.
ChatGPT alone handles billions of prompts every single day.
No consumer product in history has been adopted faster. And yet, in Jamie's words, 99% of us don't really know what we're doing with these tools. His bigger point landed harder. We keep framing AI as an IT issue. It isn't. It's about language, psychology, framing problems, and thinking critically. Things everyone can do. That reframe set the tone for the whole morning.
The experiment that flipped build vs buy.
Then came the moment I'd been waiting for. Our founder and Chief Scientist, Stuart Harris, joined Karl Hoods CBE (Group CDIO, DSIT and DESNZ) and Tom Kelly (Director of Software Engineering, UK Parliament) on stage, with Mark Chillingworth facilitating.
Stuart's first job was in 1988, when a 40-megabyte disk drive was the size of a washing machine. He's seen every wave since. He told the room this is the most exciting moment of his career. And here's why:
For 16 years, Red Badger advised our clients that if something off the shelf fits, it's cheaper than building it. The most expensive part of any system was always the software engineering. That equation has now flipped on its head.
Five days before the event, he started an experiment: replace one of Red Badger's own SaaS subscriptions with something built in-house. Not vibe coding but agent engineering, done properly. Four days later he had an MVP in production. Single sign-on. Event-sourced backend. Infrastructure as code. GitOps deployment.
The sort of thing we'd build for any client, built in just four days.
His phrase stuck with me: software becomes soft again. Malleable. Bendable to exactly what you need, rather than a generic product you contort your processes around.
For public sector leaders sitting on decades of legacy systems and expensive licences, that's not a technical curiosity. That's a strategic opening.
Nobody was selling a fantasy.
What made this panel work was the pushback.
Tom Kelly drew the line clearly: nobody should throw their finance system in the bin and work it out with an AI agent. Some problems are solved, complicated, and catastrophic to get wrong. Buy those.
He'd also just found a promising internal tool at Parliament, built on someone's laptop, manually deployed, with an API key sitting in plain text in the interface. Democratised building comes with pitfalls.
Karl Hoods put it plainly: this isn't a technology problem, it's a people problem. His teams spend their time on systems thinking and product mindset, asking why before what. And his warning was the one I'd pin on every office wall: governance has to become as agile as delivery. Speed without guardrails just gives everyone a faster way of being let down.
Quality was Stuart's answer to that. The agents are now remarkably good at writing code, but the craft lies in constraining them. Lauguages like Rust push back against the model's creativity, so only high-quality software gets through. We've spent decades accepting buggy software because humans wrote it. That bar is rising, fast.
Curiosity is the skill
The second panel brought in Avril Chester, former CTO of the Royal Pharmaceutical Society, and Dominic Hilleard, founder of Talent Leaders, to tackle the people's question.
One word kept coming back: curiosity.
Avril's advice: be brave and curious. Context is queen, keep asking why, of your stakeholders, your boss, your peers.
Dominic's line was the one delegates quoted back to me afterwards: AI won't take your job, but people using AI well might.
And Stuart offered the answer to the anxiety in the room. As a consultancy of builders, you'd think AI would be an existential threat to Red Badger.
The opposite is true. We're busier than ever and hiring faster than we ever have. Jobs aren't disappearing, they're shifting up, away from the weeds and towards systems-wide thinking.
What I took away
At the end, I asked people what they'd take home.
One delegate said curiosity being the most important skill was the thing she'd carry back. Another said simply: don't be scared of it, immerse yourself, make it work for you.
For me, it was this. The playbook genuinely has shifted more in the past six months than in the previous five years. But the fundamentals our panellists kept returning to, outcome-led, product-focused, user-centric, never technology-led, haven't moved an inch. The tools changed. The principles didn't.
That's the new playbook.
If you're a public sector digital leader wrestling with build vs buy, AI adoption or team capability, get in touch, we'd love to continue the conversation.
And again, a big thanks to Jamie Bartlett, Avril Chester, Karl Hoods CBE, Tom Kelly, Dominic Hilleard - FIRP, Mark Chillingworth and our very own Stuart Harris.