Should we build this instead of paying for it?

AI assistance has made the first version of an internal tool dramatically cheaper. It has not made the second year cheaper, and that is where this decision is won or lost.

This will not give you a verdict. It does not know what your time is worth or how good your developer is. What it will do is put the maintenance cost next to the subscription cost — which is the half people leave out — and tell you plainly where building is simply the wrong answer.

1. What would you be replacing?

2. The numbers

What it costs today ($/month) People who would use it Hourly cost of whoever maintains it ($) Hours to build the first version

3. What makes it harder than it looks

It has to talk to systems we do not control Someone is blocked from working if it breaks People outside the company would use it It holds data we would have to report a breach of One person would build it, and only they would understand it Nobody is formally responsible for maintaining it afterwards The rules or formats it depends on change without warning