I go through your product one use case at a time, decide what AI should do and what has to stay exact, then build it into the software you already have.
| What customers ask for | Approach | Why |
|---|---|---|
| Draft a reply to a support ticket | AI | A person checks it, and good enough is useful |
| Calculate an interest or leasing rate | Algorithm | It has to be right every time |
| Ask for a report in plain language | Both | AI reads the question, tested code produces the numbers |
| “Something with AI” for the sales deck | Not yet | There's no use case to build yet, so find that first |
Illustrative examples of how the decision works in practice.
The request reaches product and engineering without a use case. Building something just to tick the box costs money and helps nobody.
Find what's behind it
What are customers actually trying to get done, and where do they lose time today?
Decide each use case
AI, algorithm, both, or not worth doing, with the reasoning written down.
Give customers an answer
A concrete roadmap you can stand behind, including what you won't build and why.
Yes, but a report that looks right and contains a wrong number is worse than no report. So the work gets split.
Understands the question
“Contracts ending next quarter, by region” becomes a defined report request.
Produces the numbers
Tested queries on your existing database, respecting who may see what.
Explains the result
A readable summary, while the figures stay exactly as calculated.
Step 1
Fixed price. You get a written decision for each use case.
Step 2
I build the agreed use cases into your existing software.
Optional
A monthly arrangement, if you want a sparring partner afterwards.
Write to me about what they are asking for – your message comes straight to me.