
We've never written code faster. Are we building better software?
As AI accelerates software development, are we measuring productivity in the right way? Insights from AWS Community Summit Manchester on shifting engineering bottlenecks, evolving team structures and what the future holds for software engineering.
On October 1st, I found myself on stage at AWS Community Summit Manchester, alongside Sheen Brisals and David Anderson, questioning the future of the very profession that brought us there.
Our opening keynote, From Compilers to Copilots: The Evolution of Software and the AI-Powered Future, took the audience on a journey through three eras of software engineering. Sheen explored the past, I tackled the present, and David challenged us to rethink the future.
And the present is particularly interesting.
The productivity paradox
AI has made it remarkably easy to turn an idea into working code. As someone building products with AI, I find that incredibly exciting.
But knowing what is worth building and making it work reliably in production are entirely different challenges.
A 2025 METR study found that experienced developers believed AI made them faster, while in that particular experiment, they actually took 19% longer to complete their tasks.
AI capabilities have improved considerably since then, but another interesting contradiction has emerged.
According to Lookahead's 2026 research , 92% of respondents reported faster delivery, yet 68% experienced greater workload and pressure.
It's a pattern I recognise from many conversations with engineering leaders.
We're moving faster, but the expectations of what we should deliver are growing just as quickly.
It raises an interesting question: are we becoming more productive, or simply moving the bottleneck elsewhere, while expectations have increased everywhere?
When code is no longer the bottleneck
Throughout the history of software engineering, we've continuously found ways to make building software easier. Compilers, frameworks, cloud computing and serverless architectures have all allowed us to focus less on implementation details and more on solving problems.
AI is another chapter in that evolution. But this time, the implications go beyond technology.
Imagine doubling the speed at which your engineering team produces code without changing how you review, test or deploy it. Have you doubled your productivity, or doubled the pressure on the rest of your organization?
This is where I think engineering leaders need to pay particular attention
.
Many of our processes, organizational structures and productivity metrics were designed for a world where writing code was expensive. If that constraint changes, surely the way we organize engineering needs to change with it.
Instead of asking how much more code our engineers can produce, we should be asking how quickly we can turn ideas into measurable customer value without compromising reliability or burning out our teams.
David's closing perspective on moving from a software factory to a software garden captured this beautifully. Instead of optimizing for how much software we produce, perhaps we should focus on how well our software evolves, adapts and serves its purpose.
The opportunity with AI isn't simply to do yesterday's work faster. It's to rethink what work is worth doing in the first place.
A conversation worth having
What made Manchester particularly valuable was that these questions weren't being explored in isolation.
With over 25 talks and workshops, ComSum brought together practitioners experimenting with new technologies, operating complex systems and navigating many of the same challenges.
In an industry currently drowning in AI predictions, there's something refreshing about spending a day with people who are actually building things.
It's one of the reasons I've remained involved in the AWS community for so many years. Technology moves too quickly for any of us to figure it all out alone.
I don't believe we're witnessing the end of software engineering. But we might be witnessing the end of our obsession with measuring engineering through the production of code.
And frankly, I think that's long overdue.
A big thank you to Sheen and David for sharing the stage, the ComSum organizers and volunteers for bringing this community together, and AWS for supporting my participation.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article