A pattern I keep meeting: the system gives good answers, it scores well on the tests, and after two months almost nobody is opening it.
Almost every time, the reason is that using it means leaving the application the person lives in.
The cost of one more window
An engineer works in AutoCAD and Revit, the drawing software. A recruiter works in the system that holds all the CVs. A salesperson works in WhatsApp, whatever the sales software vendor would prefer. Asking any of them to open a separate tool costs a few seconds and breaks their concentration, and that is enough to lose them.
We published 143 tools that plug straight into AutoCAD and Revit, free for anyone to use or change, for exactly this reason. Not because it is technically interesting, although it is, but because software that does the repetitive drawing work and saves 100 to 200 hours a project only saves those hours if somebody actually runs it.
The same logic points outward
We built a sales system in which the software itself asks the first round of qualifying questions and does the chasing, inside WhatsApp and Telegram, and writes what it learns straight into the customer record so nobody retypes anything. Customers were never going to log into a website to update their own enquiry.
For a tailoring business we send order updates over WhatsApp. Nobody wrote that down as a requirement. It is simply where the conversation was already taking place.
What this costs you
Building inside somebody else's application is harder than building your own website. The connections you get to work with are worse, installing it is stranger, and you inherit whatever limits the host software imposes.
Worth it anyway. A system used by 150 people daily is worth more than a better one used by nine.
How to catch it early
In those first two weeks, ask where the person doing this work spends their day. If the answer is a piece of software nobody has budgeted to connect to, that is the problem to solve before anything else gets designed.
Working through this on a live programme?
A 45-minute call with the engineer who would run the work. We will tell you whether AI is the answer, including when it is not.