AI is changing how your teams design, build, test and ship software. That much isn’t in question. What’s harder to pin down is what responsible AI adoption actually looks like once the hype settles and your engineers are back at their desks.
That’s the question Codurance set out to answer with Matillion, one of Manchester’s standout technology success stories, and the Software Crafters Manchester community. Together, we hosted ‘AI-Powered Engineering Lightning Talks’, bringing together some of the region’s leading technology experts for an evening of short talks and honest discussion about what is really happening inside engineering teams right now, rather than the “30x engineering” claims you see on LinkedIn.
This article pulls together the key themes from the evening, written for leaders who need to make real decisions about AI adoption, and for the engineers who are living with those decisions day to day. It builds on themes we’ve explored before in our article, The 10x Engineering Reality: Beyond the AI Hype to Proven ROI.
Every engineering leader in the room agreed on one thing: speed alone isn’t the win condition anymore. The real opportunity with AI is building software that’s faster to deliver, easier to understand, safer to change and more valuable over time.
That’s a different bar to clear than “ship it quickly”. It means your teams need guardrails, not just tools. Craftsmanship, shared understanding and maintainable code matter more in the age of AI, not less, because AI can help you move fast in the wrong direction just as easily as the right one.
Yes, and this was one of the most direct themes of the evening. As AI tools become more embedded in day-to-day engineering work, they bring real convenience, but also new exposure. Code quality, data privacy, vulnerabilities, and over-reliance on generated outputs were all flagged as growing concerns. This lines up with what the wider industry is finding too: OWASP’s Gen AI Security Project1 has been tracking a growing set of security risks specific to AI-generated and AI-assisted code, from insecure patterns baked into training data through to weaker review discipline as teams move faster.
The takeaway for you as a leader isn’t to slow down AI adoption out of fear. It’s to make sure your teams have the practices in place to catch problems before they reach production, whether that’s stronger code review, better testing discipline, or simply teaching engineers to question AI output rather than accept it by default.
AI genuinely creates efficiencies, but it changes what “good” looks like for an engineer. Writing code quickly is no longer the differentiator it used to be. What matters now is:
If you’re planning your team’s development this year, this is the gap worth closing. Engineers who can direct and interrogate AI will outperform engineers who simply prompt it and move on.
No, but it changes what that human is doing. The role is shifting away from manually writing every line of code and towards guarding good practice: quality, security, architecture and maintainability. This is exactly where software craftsmanship becomes more important, not less, a point our Co-founder and Group CEO Sandro Mancuso explores in more depth in Software Craftsmanship in the AI Era.
Think of it less as AI replacing engineers and more as AI raising the bar for what a good engineer needs to protect.
They have to. Several speakers made the point that AI can help teams move faster, but speed without discipline creates more technical debt, not less. Strong principles around testing, clean code, documentation, observability and governance are what let your teams scale what they build rather than just accumulate debt. This mirrors a central finding from Google Cloud’s DORA State of AI-assisted Software Development report2: AI acts as an amplifier, magnifying whatever strengths or weaknesses already exist in a team’s engineering practices, rather than fixing them on its own.
If your organisation is under pressure to show fast AI wins, this is worth raising directly with your leadership: speed you can’t sustain isn’t really speed, it’s borrowed time.
Both, and this was one of the more sobering points of the evening. Legacy systems will keep breaking, becoming harder to maintain, and creating barriers to growth, particularly where legacy technology underpins critical operations. Software engineers remain the people who fix, modernise and improve those systems.
AI programmes will break too. Models drift, integrations fail, and outputs become unreliable over time. Someone still needs to monitor, maintain and improve those systems, and that someone is your engineering team.
The message from the evening was clear: AI doesn’t remove the need for strong engineering capability. It raises the bar. You still need skilled engineers who can apply judgement, protect quality, manage risk, and make sure AI is used in a way that’s safe, scalable and commercially valuable.
If this article raised questions your organisation hasn’t fully answered yet, you’re not alone. Most engineering leaders we speak to are somewhere between “we’ve started using AI” and “we’re not sure it’s being used safely”.
Codurance helps engineering teams build guardrails that make AI adoption sustainable: strong testing practices, clean architecture, code review discipline and the craftsmanship principles that keep systems maintainable as they scale. Whether you need an assessment of how AI is currently being used across your teams, or hands-on support building the practices to use it responsibly, our software craftspeople work alongside your engineers to get you there.
Want to talk through what responsible AI adoption looks like for your team? Get in touch with Codurance to start the conversation.
Does AI reduce the need for skilled software engineers?
No. It changes what skilled engineering looks like. Engineers now need to review, question and direct AI output, alongside the traditional skills of good architecture, testing and maintainability.
What’s the biggest risk of AI-powered engineering right now?
Moving fast without the guardrails to match, which shows up as security exposure, code quality issues and technical debt surfaces later.
Is software craftsmanship still relevant in an AI-driven world?
More relevant than ever. As AI takes on more of the writing, the human role shifts towards protecting quality, security and long-term maintainability, which is the heart of software craftsmanship.
What skills should engineers focus on developing as AI adoption grows?
Writing better prompts, reviewing AI-generated code critically, and knowing when to challenge AI output rather than accept it at face value. These matter more now than raw coding speed.
How can engineering teams stop AI from creating technical debt?
By keeping strong principles around testing, clean code, documentation, observability and governance in place. AI can help teams move faster, but without this discipline, that speed just becomes debt to pay off later.
Who is responsible for maintaining AI systems once they’re in production?
Software engineers. Like any other system, AI programmes drift, integrations fail, and outputs become unreliable over time, so they need ongoing monitoring, maintenance and improvement from skilled engineers.
Lee Sanderson is a Principal Software Craftsperson at Codurance. He has over 30 years of experience in software development, spanning roles from Developer to hands-on CTO, and is a polyglot programmer with deep experience across modern technologies. He enjoys helping teams move faster by streamlining development through XP, Agile practices, and continuous delivery, while maintaining clean architecture and high-quality code. He also enjoys spending time exploring how AI will shape the future of software engineering.
1. OWASP’s Gen AI Security Project
2. DORA State of AI-assisted Software Development report, published in 2025