Blog
Ask your AI: brief me
One question replaces the six an assistant used to ask to work out how yesterday went — and the answer says what it could not measure instead of filling the gap with a zero.
"Brief me — how did yesterday go, and is anything broken?" is now one question with one answer. It used to be six or seven: what accounts do I have, how did they do, what did they earn, what were the best posts, what is scheduled, did anything fail. An assistant could ask all of those and stitch them together, and the stitching is where it went wrong — each question resolved its own dates, and two slightly different windows is how a summary ends up disagreeing with itself.
What it answers with
This is recorded, not written. It came out of a demonstration workspace — invented accounts, invented posts — by running the question through the same door your own assistant comes in at.
You: Brief me — how did yesterday go, and is anything broken? It: On 2026-08-25, Ledger Lane took 104,414 views against 68,151 for the median of the 28 days before it — up 53.2%. Daily Brew earned $15.52 that day. 3 of your accounts have no earnings figure at all — they are not in a monetization programme, so that is an absence rather than a zero, and I have left them out of the total instead of averaging them in. Nothing is scheduled in the next 24 hours.
The third line is the one worth reading twice.
Nothing measured is not nothing earned
Three of those accounts are on a platform that pays no per-view revenue at all. The easy thing — the thing almost every dashboard does — is to add them in as zero, because zero is a number and numbers fit in a total. That quietly drags every rate down and reports a figure that measures nothing.
So the brief says which accounts had nothing measured and why, and leaves them out. The same rule runs everywhere: a day that is outside what has been synced is not a quiet day, and a page that has never been monetized did not earn nothing. Posts stays a real zero, because "you published nothing" is a finding you can act on.
It tells you when it does not know
The same discipline shows up in the question everybody asks second:
You: When should I be posting? Use my own history. It: No posting slot has cleared the evidence bar yet. A weekday-and-hour slot needs at least 3 of this account's own posts inside the last 180 days before it is worth stating. This is a statement about how much history there is, NOT about the account's times being bad.
An empty list would have read as "there are no good times". A confident answer off two posts would have been worse. Saying which bar has not been cleared, and how far off it is, is the only version of that answer somebody can act on.
Where to try it
Every screen in SupahPost now carries three questions it can answer — above your numbers, beside a post, on the calendar, and next to an account that has stopped working. Press one and the assistant opens with the question written out and not sent, so you can change it first. The same sentence works in Claude, ChatGPT or Cursor once you have connected them, because both are asking the same tools.
The whole list of what you can ask is on Your AI, and the setup for each assistant is in connecting your AI.
The transcripts above were recorded from a demonstration workspace and the connection steps for each assistant were checked 2026-09-08.