Nothing to hand? Load the — five accomplishments, most with a number behind them — the , where the notes are thin and the honest output flags what's missing instead of inventing it, or the , a nine-person roster whose Exceeds bucket runs well over the target band.
Paste your notes — the scan is free
No upload, no AI: the scan reads your notes (or your pasted roster, in calibration mode) in your browser and lists what it mechanically found — how many accomplishment-like lines it counted, how many carry no number or measurable result, whether goals from last cycle were mentioned, and for a roster, the rating distribution against your target bands. Each flag explains why it matters. This part costs nothing and happens while you type.
The AI writes — this is the metered part
A senior HR business partner's pass over everything you pasted: for a self-assessment, each accomplishment gets its situation, contribution and impact plus the source quote it traces to; for a manager review, a rating recommendation with justification, strengths and development areas with specific examples, and a compensation direction that is never an invented number; for calibration, every roster row carried through faithfully with a computed rating distribution, discussion points and promotion candidates grounded in the notes column. Anything the notes don't support becomes an open question instead of a fabricated fact. A worst-case amount is reserved before the run and only what the run uses is charged.
Ship the document
Copy it, download it as Markdown or JSON, or print it. For calibration, the roster table also exports as CSV. Past reviews are saved to your SkillSafe account, so they follow you across devices: when a mid-cycle check-in adds new material, restore the last run, add the new notes, and re-run to see how the picture changed.
A derived work of @anthropics/performance-review (Apache-2.0).