ChatGPT vs a custom AI tool: which does your business need?
You don't always need something built. Sometimes a £20 subscription is the right answer. Here's how to tell the difference before you spend anything.
I build custom AI tools for a living, and I'm about to talk some of you out of buying one. Because half the businesses asking me for a custom build would be better served by a twenty-quid subscription and an afternoon of setup. Knowing which half you're in saves you a lot of money.
When ChatGPT (or similar) is enough
Reach for the off-the-shelf tool when the work is general and the person using it can check the answer. Drafting emails. Summarising a long document. Rewriting a clunky paragraph. Brainstorming. Getting unstuck on a first draft. This is genuinely useful, available today, and cheap. If that's your need, you don't have a software project. You have a subscription and a bit of training to do.
The tell: the task changes every time, a human eyes every output, and nothing needs to know your specific rules or connect to your systems.
When you need something built
You cross into custom territory when the tool needs to know things a general model can't, or do the same job the same way every time. It has to know your exact products, prices, rules or lender panel. It has to plug into your data so nobody's copying and pasting. It has to do a repetitive job identically, for the whole team, without someone re-explaining the context each time.
The tell: you keep pasting the same background into ChatGPT to get a usable answer, and you're doing it several times a day, across several people. That copy-paste is the thing a build removes.
The middle option most people miss
It's rarely all-or-nothing. Often the right move is a thin custom layer on top of an existing model: you're not building a new brain, you're giving a very capable one your business's knowledge and connecting it to your workflow. That's far cheaper and quicker than owners expect, because the hard part — the intelligence — already exists. You're just pointing it at your problem.
How to decide
Try the cheap thing first. Live in ChatGPT for a fortnight on the job you have in mind. If it solves it, brilliant, you're done and you've spent almost nothing. If you find yourself fighting it — re-pasting the same context, correcting the same mistakes, wishing it just knew your numbers — that friction is your spec. It's telling you exactly what a custom tool would need to do, and exactly why it'd be worth it.
Not sure which side of the line you're on?
Answer three questions and Emmanuel will tell you honestly: a subscription, a build, or nothing yet.
Diagnose the drag