Calibration
Confidence, measured against what happened.
Everyone thinks they are about right. Calibration is the arithmetic that checks it, built only from cases you have actually closed.
How it is calculated
Closed cases are bucketed by the confidence you claimed. Within each bucket, the share of predictions that held is compared against the confidence you wrote. If your 80% bucket holds 55% of the time, you are overconfident by 25 points — and now you know by how much rather than in the abstract.
Which objections mattered
Each check-in records the objection that turned out to matter. Aggregated across cases, that shows the failure modes you habitually wave away — the cost you never count, the dependency you always assume will hold.
How much history it uses
Calibration is only as good as the record behind it. Free accounts keep the five most recent cases, which is enough to see the idea working. Advocate and Counsel keep every case, so the numbers keep improving instead of resetting.
Start with the free tier — no card, five credits to spend.
Back to the Desk · Terms · Refunds · Privacy