We’d wanted to rebuild Strategy Connect for years.
Strategy Connect was a platform we’d built and maintained over many years.
The software had evolved, the requirements had changed, and our understanding of the product had grown considerably.
We knew there were things we’d do differently if we were starting again.
But rebuilding a working platform is expensive. Months of development, significant investment and the risk of disrupting something customers already depend on.
There was always something more immediately valuable to work on.
The rebuild made technical sense. It just didn’t make economic sense.
Then the client suggested we try something different.
Aidan, the owner of Strategy Connect, knew I’d been experimenting with AI-assisted development.
We started discussing whether the technology had progressed enough to change the economics of a rebuild.
Rather than committing to a conventional development project, we could try building a new foundation using AI-assisted tools and see how far we got.
We had years of product knowledge, a clear understanding of the business and plenty of ideas about what we’d change.
What we didn’t know was how much faster this new workflow could make us.
The cost of finding out had become small enough to give it a go.
The biggest change wasn’t the code. It was the feedback loop.
Traditionally, turning an architectural idea into working software takes time.
You design the approach, implement it, run it, review the result, make changes and repeat.
AI-assisted development compressed that cycle considerably.
I could describe an architectural direction, work through an implementation with the AI, run the software, review what it had produced and immediately refine it.
The AI handled much of the implementation work, while I concentrated on architecture, product behaviour, integration and deciding whether the result was actually any good.
It wasn’t a case of typing a prompt and walking away.
There were mistakes, corrections, technical decisions and plenty of iteration.
But the pace was completely different.
I was spending more time deciding what the software should be and less time translating those decisions into code.
That’s when the workflow really started to click for me.
Work we’d expected to take months was suddenly achievable in weeks.
We made progress far faster than I would have expected using our previous approach.
A new technical foundation that had been difficult to justify commercially became practical to build and evaluate.
And importantly, we weren’t starting from nothing.
We had years of understanding the product, its users, its requirements and the problems we wanted to solve.
That knowledge mattered enormously.
AI didn’t invent the product or make the architectural decisions for us. It accelerated the process of turning those decisions into working software.
The experience changed my expectations of what a small, experienced team could achieve.
It also changed how I thought about the economics of software development.
Then Aidan started building too.
One of the most interesting developments came after the initial rebuild.
Aidan began using AI-assisted development to make changes to the platform himself.
He isn’t a software developer. But he understands Strategy Connect, its customers and the business better than anyone.
Previously, making a change generally meant explaining what he wanted, having a developer implement it, reviewing the result and repeating the process.
Now he could participate much more directly.
That doesn’t mean every change is automatically production-ready or that engineering oversight becomes unnecessary.
But it does mean someone with deep product knowledge can get much closer to the implementation.
And I find that fascinating.
AI isn’t just accelerating development. It’s changing who can participate in creating software.
It feels like a new superpower.
I’ve spent decades learning how to design and build software.
That experience hasn’t become less useful. If anything, I’ve found myself drawing on it more.
Understanding architecture, recognising problems, knowing what to test and being able to judge the quality of an implementation all become important when software can be generated so quickly.
The difference is the speed at which I can put that experience to work.
An idea no longer has to wait until I’ve found the time to implement every detail myself.
I can explore an approach, build a working version, test it and change direction while the thinking is still fresh.
It’s a new level of leverage for someone who already knows how to build software.
And it’s made building exciting in a way I haven’t felt for a long time.
Speed changes what becomes worth building.
For me, the most interesting consequence isn’t simply delivering the same software faster.
It’s what happens to ideas that previously weren’t worth pursuing.
A rebuild that was too expensive.
A feature that would take too long to validate.
An experiment that needed a team before we could even find out whether it worked.
When the cost of turning an idea into working software falls dramatically, the calculation changes.
You can test more ideas. Explore more alternatives. Discover mistakes earlier. Spend less time debating what might work and more time finding out.
Of course, faster implementation doesn’t automatically mean better products.
You still need to understand the problem, make good decisions, test what you’ve built and put it in front of real users.
But now we have a much more powerful way of doing that.
And that opens up possibilities I find genuinely exciting.
I don’t want to go back.
I still enjoy programming. I still care about architecture and engineering.
But I don’t miss spending days implementing something that I can now explore in hours.
AI-assisted development has become part of how I work, alongside the engineering practices and product judgement I’ve developed over my career.
I’m still learning where it works well, where it struggles and how to use it responsibly.
But the Strategy Connect experience changed my perspective.
We haven’t made software engineering irrelevant.
We’ve given ourselves a much more powerful way to practise it.
And I’m only beginning to explore what that makes possible.