The gap the reader fills for you
Almost every message we send at work leaves a little room, and we have all learned to live with it. "Let's circle back next week." "Can you tighten this up?" The words leave a gap, and the reader fills it with their own assumption about what we meant.
Most of the time we do not stop to check. We move ahead on the reading that feels most likely and trust it matches theirs. It is not carelessness — it is how communication works, and most days it holds.
Until it does not. The proposal a client reads the opposite way. The note that sends half the team in the wrong direction. The board deck where one vague line quietly loses the room. The failure is rarely a bad idea. It is a sentence that could be read two ways, and the reader chose the one you did not intend.
The habit, and the exact ask
Before anything important goes out, ask AI to find the ambiguity for you. Not to rewrite it — to review it.
"Show me every place this could be read more than one way, and which reading is most likely to cause a problem. Don't rewrite it."
What comes back is usually three or four spots where you had assumed a shared context that was not actually on the page.
Then triage what comes back, because most of it does not matter
A review like this over-reports. It will flag genuinely ambiguous phrases that no reader would ever trip on, and if you fix all of them you end up with a document written defensively, which is its own kind of unclear.
One test sorts them. For each flagged phrase, ask what each reading would cause someone to do differently. If both readings lead to the same action, leave it — that ambiguity is free. If they lead to different actions, or to one person acting and another waiting, that is the sentence worth ten more seconds.
Usually one or two of the four survive that test. Those are the ones that would have cost you the week.
Why this matters past the single email
It is AI used as a reviewer, not an author. The judgment stays yours — the machine just catches the gaps you cannot see, precisely because you already know what you meant. That is the whole difference between handing work to AI and keeping ownership of it, and it is a habit an entire team can adopt without a single new tool.
Worth one pass this week, on something that matters.
Where do unchecked assumptions already flow through your team's work? We redesign how the work gets done so judgment stays with your people, not the tool: See how →
Keep reading: Hold up the MIRROR · Navigation, not generation
This is the thinking behind a Trusted AI Culture.
The practice of turning AI on your own draft to interrogate it, rather than to produce it, is the MIRROR method — Dr. Jules White, Vanderbilt University.