Bespoke software for small businesses: now the shoe will always fit

Most owners we speak to have priced custom software at least once. Someone quoted a number, attached a timeline, and the decision made itself. So you bought the off-the-peg product instead, and then you did the thing everybody does: you changed how your business worked to suit the software.

That is the part nobody puts in the business case. The licence fee is visible. The workaround is not. The spreadsheet that sits beside the system because the system will not hold that one field. The stage in your process that exists only because the tool needs it. The job someone does every Friday to move data from the thing that captured it to the thing that needs it.

You bought a shoe that was made for the average foot of several thousand businesses, and then you learned to walk differently.

Software for thousands of users is not software for your business

An off-the-peg product has to work for everyone who buys it. That is not a criticism, it is the economics: build once, sell many, and the only way that maths works is to serve the middle of the market. Every feature is a compromise between customers who want different things, and the compromise is settled by whoever there are most of. You are not in that group. Nobody is. There is no average business.

What we are talking about here is different, and it is smaller than people expect. Not a platform. Not a system that replaces your accounts package. A small piece of software built around one action your business actually takes: the thing you do forty times a week that no product on the market does the way you do it.

The intake that has to check three conditions before it becomes a job. The pricing rule that lives in your head and gets applied inconsistently by whoever is on. The handover between two people that currently happens by shouting. These are not features anybody is going to build for you, because you are one customer. They are also, usually, where the time goes.

What actually changed

The cost of building software has fallen sharply, and it has fallen most at exactly the size that used to be uneconomic. A small, well-specified tool that does one job for one business used to fail on arithmetic: the engineering time to build it cost more than the problem cost you. For a big build that maths still deserves scrutiny. For a small one it has quietly inverted.

We run our own businesses on this. ScottishWill is a will-writing service where the intake, the payment matching, the document production and the quality checks are handled by software written for that specific pipeline, not bought. One person carries the admin load of four. On the property side, receipts are read, matched to the accounting record and filed without anyone opening a folder. None of that is a product you can buy. All of it is small.

The point is not that we are clever. It is that these builds are now the size of a decision an owner can actually make, rather than a capital project that needs a board behind it.

The honest limit

This does not work everywhere, and the sharpest version of the caveat comes from someone with every reason to oversell it. Dharmesh Shah co-founded HubSpot and now builds AI agents for a living. Speaking on the My First Million podcast in September 2025, he drew the line like this: AI-built software is very good when you are building for yourself, when it has a short lifespan, when it is disposable, or when it is for a small team and it is not a disaster if it stops. Violate those conditions, he said, and you will be painted into a corner.

He is right, and it matters more than the hype does. Your accounting system, your system of record, the thing that holds data you file with a regulator: leave it alone. Those earned their place by being boring and dependable, and nothing about this changes that.

But read his list again, because most small business problems sit inside it rather than outside it. Small scale. Specific. Owned by you. And there is a second implication people miss: if software is cheap to build, it no longer has to last ten years. It has to fit how you work this year. When the business changes, you change the tool, instead of the tool quietly deciding how the business runs.

That is the real shift. Bespoke used to mean expensive and permanent, which is a bad combination for a business that is still moving. Cheap and disposable is a much better one.

Where to start

Not with software. Start by finding the workarounds, because they are the evidence. Every spreadsheet that shadows a system, every recurring manual step, every rule that lives in one person's head is a place where the shoe does not fit. Most owners can list five without thinking hard, and are so used to them that they no longer register as problems.

Then work out what each one costs in hours and errors, and how much of it is a genuinely specific action versus something a product already does well. Some of the answers will be to configure what you have properly. Some will be an AI employee handling the reading, chasing and filing. A few will be worth building, and those few are usually where the time has been going all along.

That is what our AI Operations Day is: £750 per day for a working day inside the business, watching where the work waits and what gets repeated, followed by a written Opportunity Map with a straight answer about which of it is worth automating, building or leaving alone. If you priced custom software once and walked away from the number, it is worth pricing again. The answer has changed.