The Questions to Ask Before You Pay Anyone for a 'Custom AI Solution'

Every agency in the country has an "AI solution" on offer right now. Some of that is real automation that will save a business hours a week. A lot of it is a wrapper around a chatbot API with a markup on top. Before you sign anything, a handful of direct questions will tell you which one you're being sold.

"What exactly does this replace?"

A real AI project has a specific, current process it's replacing: someone manually re-keying enquiry forms, a report that takes half a day to assemble, a set of customer questions answered the same way every time. If the answer to "what does this replace?" is vague, or the pitch starts from the technology instead of your process, that's worth pausing on.

"What happens when it gets something wrong?"

Every AI system gets things wrong sometimes. The question isn't whether that happens, it's what's in place when it does: does a person review anything that matters before it goes out, or is the system trusted to act unsupervised on things like customer communication or financial data? A serious proposal has a clear answer to this before you ask it.

"Is this actually built for us, or reconfigured for us?"

There's a real difference between a system designed around how your business runs and a generic template with your logo added. Neither is automatically wrong, but you should know which one you're paying "custom" prices for, and what happens if you want to change how it works six months in.

"What's the smallest version of this we could try first?"

The AI projects that hold up are usually narrow: one process, automated properly, expanded once it's proven itself. If the proposal in front of you is "AI across the business" with no smaller starting point, that's usually a sign it hasn't been thought through past the pitch deck.

Asking these four questions won't make you an AI expert, and it doesn't need to. It puts the burden of a clear answer back where it belongs: on whoever's asking you to pay for it.