Writing a careful public reply

A resident wrote in because a rider had heard that one of our services might change hands, and the rider was worried. Nothing had been decided. That is exactly the kind of message where a careless sentence becomes a commitment.

What I asked

I asked for a reply that was honest and reassuring: the City has looked at options, nothing has been decided, and any change would go through the City's normal procurement process and the City Council.

What came back

Two versions, a short one and a fuller one. I picked the short one.

The first drafts also contained a mistake. Claude wrote as if a provider had already been chosen and a purchasing method selected. Neither was true. I said so, and it corrected the draft at once. I mention this because it is the most important lesson in the post: the AI will sometimes treat something you are exploring as something you have decided. In a public reply, that is the error that matters most.

It then drafted a cover note sending the reply to our public information officer for review before anything went out. That is our process, and the draft followed it.

A second example

A community group asked my director for a guest speaker, and she passed the request to me. Claude's first draft wrote to the group as though they had contacted me. They had not. They had never heard of me. The second draft introduced me, explained how the request reached me, and then said yes.

What I take from this

Tell it who is writing to whom, and what each person already knows. That context is the difference between a reply that sounds right and one that reads as slightly off.

Where it falls short

It does not know what has been decided and what has not. You do. Read every public-facing sentence as if it will be quoted back to you.

---

Part of the series 27 ways I actually use AI. Claude drafted this post from the record of our work together.