Why one developer can now do the work of a team — and what that means for your budget

Not long ago, getting a piece of custom software built meant hiring a team. A project manager, a couple of front-end developers, a back-end developer, someone for the database, a designer, a tester — and the budget that comes with feeding all of them for months. That's why so much small-business software either never got built or got handed to an offshore shop and came back not quite right. The maths simply didn't work for a business turning over a few hundred thousand a year.

Over the last couple of years that maths has quietly changed. Not because software got less complicated, but because the tools a developer uses got dramatically better. I want to be straight about what's actually shifted, because there's a lot of hype around it — and the honest version is more useful to you than the breathless one.

What used to take a team, and why

A traditional build needed a team for two reasons. First, the sheer volume of work: writing every screen, every form, every bit of plumbing between the database and the page took human hours, and there were only so many in a day. Second, breadth of skill: front-end, back-end, infrastructure, testing and design were genuinely separate specialisms, and no one person was fast at all of them. So you hired one of each, added a manager to coordinate them, and absorbed the overhead of everyone talking to everyone else.

Most of that cost was never the thinking. It was the typing, the coordination, and the handoffs — work that had to happen but didn't, on its own, make your software any better.

What AI actually changed

AI coding tools haven't replaced the judgement in software — they've collapsed the volume-of-work problem and softened the breadth-of-skill one. The repetitive parts that used to eat most of the hours now take a fraction of the time:

• The boilerplate. The forms, the screens, the standard plumbing between database and page — written in minutes instead of days.
• The breadth. A senior engineer who's strong in one area can now move confidently across the whole stack, with AI filling the gaps that used to need a specialist.
• The grind. Tests, migrations, refactors, the unglamorous maintenance work — drafted in seconds and checked, rather than typed by hand.
• The research. Working out how an unfamiliar API or library behaves used to mean hours of reading. Now it's a conversation.

Add those up and a single experienced developer can now ship, in a few weeks, what genuinely used to take a small team a few months. That's not a marketing claim — it's the daily reality of how I work now versus how the same work was done five years ago.

What AI didn't change — and this is the important part

Here's where I part company with the hype. AI makes a good engineer much faster. It does not make software build itself, and it does not replace the parts that actually decide whether your project succeeds:

• Knowing what to build. Turning 'my bookings are a mess' into the right system is judgement, not typing — and getting it wrong is the expensive mistake AI can't save you from.
• Architecture and trade-offs. AI will happily write code that works today and falls over at scale. Deciding how it should be built is still a human call.
• Spotting when it's wrong. AI is confidently wrong often enough that you need someone with the experience to catch it before it ships, not after.
• Owning the outcome. When something breaks at 9pm, a tool doesn't answer. A person who built it and understands it does.

What this means for what you pay

When the volume of work collapses and one person can cover the breadth a team used to, two things follow for you. The cost of a custom build comes down, because you're no longer paying for a team's worth of hours and the management overhead of coordinating them. And the things that were never worth building before — too small to justify an agency, too specific for off-the-shelf — suddenly are. The floor for 'worth doing properly' has dropped.

That's the whole reason I can do what I do. A senior engineer working solo, using these tools well, can deliver work that genuinely used to need an agency — and charge a fraction of what an agency charges, because the cost structure underneath is a fraction of what it was. You're not getting a junior cutting corners; you're getting senior judgement made faster, with the overhead stripped out.

Why this matters now, not later

This window is new, and most small businesses haven't caught up to it yet. They still assume custom software means a five-figure agency quote and six months of waiting, so they put up with the spreadsheet held together by hand and the off-the-shelf tool that almost fits. The businesses that move now — that get the booking system, the joined-up workflow, the small custom tool built around how they actually work — get ahead while their competitors are still assuming it's out of reach.

It also changes who you should hire. When one experienced person can do the work, paying for a team — or a layer of account managers between you and whoever actually writes the code — stops making sense. You want the person doing the building to be the person you talk to. That's the point of how I'm set up: you explain your problem to me, I build it, and there's no middleman in between adding cost and losing detail.

If you've been assuming custom software is out of your budget, it may be worth a fresh look — the maths has changed more than most people realise. Tell me what's slowing your business down and I'll tell you honestly whether it's worth building, and roughly what it would take. No team, no account manager — just me.