Start with the problem.The stack can wait.

What to put in it

A first reply is only as good as the enquiry. Four things make ours worth reading.

01The decision you want to improve

Not the feature you have in mind — the call someone in your business currently makes on instinct, or too slowly, or too late. That is what tells us whether a model helps at all.

02What data already exists

Roughly what you hold, where it lives, and how far back it goes. "Three years of orders in Postgres, plus a warehouse nobody trusts" is more useful than a polished spec.

03What you have already tried

The spreadsheet, the rule of thumb, the vendor that did not work out. Knowing the baseline stops us proposing something you have already ruled out.

04Your constraint

A date, a budget ceiling, a compliance requirement, an integration that cannot break. Say it early and the recommendation will be worth reading.