The shackles are off.

Two or three years ago, if something in your business genuinely needed a piece of software, the path was well-worn and not much fun.

You'd sit down with a developer. You'd try to describe what you wanted. They'd go away and come back with something half-right. You'd refine it. They'd go away again. Three meetings in, the spec would have drifted and the estimate would have doubled. Six weeks later you'd have a working version of the thing you asked for eight weeks ago, which was already slightly not what you needed any more. And making changes? Another engagement. Another three calls. Another round of red tape.

I can remember going through this with Microsoft Dynamics. We looked at basing the business on it — genuinely looked at it — and it quietly turned into thousands of pounds of back-and-forth for something that wasn't exciting, wasn't especially functional, and wasn't especially useful at the end of it. The fatigue alone was a cost nobody put on the invoice. By the time you'd fought your way to a workable version, the appetite for improving it further was gone.

And that's the half of the story that got built. There's another half that's more interesting, which is the half that never did.

Think about the small asks. A front end to log visitors on arrival. A simple record so the warehouse knows which supplier a pallet came from. A neat little interface to the access-control system so reception can issue a temporary badge without opening three other applications. None of those were worth going to a software house for. They were too small. The engagement cost would have dwarfed the value, and anyone running a business knew it. So those asks got filed away in the category of things we'd do if the project was ten times this size, and mostly they never happened.

A whole layer of perfectly useful small software just didn't get built, because the economics didn't work.

Fast forward to now.

We build those small, useful things for clients regularly, and the engagement looks almost nothing like the one I've just described. The cost is a fraction of what it was. The loop between "here's what I want" and "here's a working version you can touch" is hours or days, not weeks. Changes are cheap, because the iteration cost has collapsed — you react to what you see on screen rather than trying to specify every edge case in advance. And because the unit cost has collapsed — a middle-tier model answering a real question costs fractions of a penny — the small asks that used to be uneconomic to build are the sweet spot. Visitor systems. Internal dashboards. Little workflow glue between two systems that don't talk natively. The stuff that would have been dismissed as overkill eighteen months ago is now exactly the shape of thing that makes sense to commission.

Three shackles have come off at once. Accessibility — because the conversation doesn't require you to be fluent in software engineering to have. Time — because the build loop is days rather than months. Cost — because what used to be a five-figure project is often a four-figure one, and what used to be a four-figure project might be an afternoon.

That combination is what makes this a genuinely different era, not a slightly faster version of the old one. It's not that developers got quicker. It's that the entire economics of "should we just build this" shifted, and the category of small, specific, useful-but-not-worth-it collapsed almost overnight.

This is the same shift I wrote about in the last post from a different angle. There, it was the emotional side — the commoditisation that wakes you up at three in the morning if you're on the supplier side of the line. This is the same commoditisation seen from the buyer's side, and from the buyer's side it reads mostly as good news. Work that used to need a specialist, or would never have been commissioned at all, is now inside reach.

It links back to the automation-over-AI point too. The headline-grabbing AI demos aren't usually where the value lives for a business your size. The value lives in joining things up, removing manual steps, making a small piece of software that takes the friction out of one specific part of the day. That's always been where the gains were — the difference is that the cost of doing it has fallen through the floor.

And it links back to starting with the boring bit. The way you find the opportunities hasn't changed — you look at where time is getting lost, where people are doing things by hand, where a workflow has three steps that used to have one. What's changed is what's now possible once you've spotted one. The answer used to be "we'll park that until we have budget for a proper project." It's now "let's build it next week."

The mental model most of us are still carrying is the old one, and it quietly filters out half the things we could ask for. The sentence "that's too small to commission" is doing a lot of work in the background that it didn't used to do, and most of it is wrong now.

So if you take one thing from this, it's that — start noticing when you're about to dismiss an ask as too small or too niche or too bespoke. Those are the ones worth raising. The shackles are off, and most businesses haven't quite let themselves believe it yet.