How it works
An afternoon to set up. Ten minutes a month to run.
The whole point is that the expensive work happens once, at the start, and the rest accumulates on its own. Here is exactly what you do, and what happens without you.
Day one · about an hour
Say what you expect
Add your reports and write down what you actually want from each of them. Not values, not a competency framework — four or five observable things, of the kind you would recognise if you saw one.
For example
- "Raises a risk while there is still time to do something about it, not in the retro."
- "Leaves a review comment that explains the reasoning, not just the objection."
- "Takes an ambiguous problem and comes back with options, not with questions."
If a blank page is the obstacle, generate a draft from the role and delete what does not apply. Editing somebody else's list is a far easier job than writing your own, and the point is that the final list is yours.
Then pick where each person is heading. It can be the obvious next role, a sideways move, or a stretch you have not told them about yet. This is what turns a pile of expectations into a direction.
Every week · two minutes, mostly theirs
Wins land as they happen
Your report writes down what they did, in the moment, in two lines: the impact, and how they got it. It takes a couple of minutes then and is genuinely impossible nine months later.
AI reads what they wrote and suggests which of your expectations it demonstrates and which skills it used. They accept the ones that fit, drop the ones that do not, and link the goal it moved if there is one. Then it lands in your inbox and you approve it — one click, on a phone, between meetings.
If your team does not write anything
Write the wins yourself. It is how most managers start, it takes under a minute each, and anything you write is approved on save. Teams usually flip round within a month or two — once somebody sees their own record being read out in a check-in, they start keeping it themselves.
Every month · ten minutes
A check-in that has already done the prep
The check-in opens with everything since the last one on it: the wins that landed, the expectations that moved, and the things you both agreed to a month ago and may otherwise never have mentioned again.
Your report says how the month went and what they want to work on. You add a grade and your own read. Agreements you make get a deadline and a name against them, and they stay open until they are closed as done, not delivered, or cancelled — three answers, because "we dropped it" and "it shipped" are different facts and a checkbox cannot tell them apart.
Agreements belong to the person rather than to the meeting, so an open one is read again by every check-in that follows. Nothing has to be copied forward, and nothing quietly disappears.
Continuously · nothing at all
The plan and the review keep themselves current
Every approved win recalculates the growth plan: readiness against the target, which expectations are still missing, which skills would unlock them. You never sit down to build one.
And because the evidence went in dated and linked, the performance review is assembled as you go rather than written at the end. At any point in the year you can open a report's page and read what they did, what it proved, and what moved because of it.
Review season · an hour, not a weekend
You read instead of writing
Open the page. It is all there, in order, with dates. Add your judgment — which is the part only you can do — and you are finished.
If your company runs its review cycle somewhere else, copy out of a document that is already written. And when you go into calibration to argue for somebody, you go in with a case rather than an adjective.
What this still asks of you
No tool manages for you. Here is the honest list of what stays your job.
Deciding what you expect
AI can draft it. Only you know which four things actually matter on your team this year.
Judging the work
Nothing here scores anybody. Whether a behaviour is consistent is a call you make, and the system only records that you made it.
Approving in roughly a week
Evidence approved a month late still counts, but the loop only feels alive if the queue stays short.
Having the conversation
A record makes a hard conversation possible and specific. It does not make it happen.
Where the loop lives
Your inbox, the check-in, and the growth plan. That is most of it.
Stop writing reviews from memory.
Set up your first report in about ten minutes. Free forever for teams of five.
No credit card. Your reports never pay.