Who owns the AI decision in a thirty-person company?

A recurring pattern: a company discusses an AI tool in March, does not decide, discusses it again in June, does not decide, and discusses it again in September. Nothing has changed except that everyone is now slightly more tired of the subject.

This is not indecision. It is an ownership gap, and it has a specific shape in companies of twenty to eighty people.

Why it lands nowhere

The decision touches IT, which does not want to own a business question. It touches operations, which does not want to own a technical one. It touches the managing director, who has eleven other things. So it gets discussed by all three and decided by none.

Larger companies solve this with a role that does not exist at thirty people. Smaller ones solve it because the owner decides everything anyway. Your size is the awkward middle.

The fix is one sentence, not a committee

Name one person who decides, and say what they decide without asking.

Something like: Anna decides which tools we use and what may go into them. Spending above two thousand francs a year comes to me. Everything else is hers.

That is the whole governance structure a thirty-person company needs. It is not a policy document, it is a sentence in a meeting, and it ends the loop permanently.

Who it should be

Not necessarily the most technical person. The right choice is usually whoever has the best view of how work actually flows through the company — often someone in operations or administration who has been there a while.

Technical ability can be borrowed. Knowing which process is actually painful cannot.

What this unlocks

Everything in this blog for the past year assumed somebody was in a position to act: to ask a vendor a question, to turn off a setting, to write the one-page policy, to ask five colleagues why they are not using the tool.

If you have read all of that and done none of it, this is probably the missing piece. September is a good month to say the sentence.