Product · Candor vs manual tracking
A spreadsheet can be edited after the fact. The record can’t.
Most firms “track” AI use in a spreadsheet or a shared doc, the real status quo. It’s better than nothing, but it proves what someone typed, not what happened, and it’s the first thing a careful examiner discounts. Here’s the difference when a carrier asks.
| Feature | Manual tracking (spreadsheet) | Candor |
|---|---|---|
| Can it be edited after the fact? | Yes, any cell, any time, no trace | No: append-only and hash-chained; any edit is detectable |
| When is use captured? | From memory, later, if someone remembers | At the point of use, in the AI tool itself |
| Client-data screening | None | Confidentiality pre-check before text reaches a model |
| Citation checks | None | Flags cited cases that may need a second look (CourtListener) |
| Attorney sign-off | A name in a cell | A recorded review tied to the verified person, append-only |
| Carrier-renewal report | Assembled by hand under pressure | Generated from the same record, on demand |
| Holds up when examined | Only as much as the honor system behind it | The chain verifies, or it doesn’t, provably |
Why Candor for law firms
The point isn’t logging. It’s a record that survives scrutiny.
The reason to keep an AI-use record at all is the day someone asks to see it: a carrier at renewal, a court, a client. A spreadsheet answers that question with “trust us.” Candor answers it with a tamper-evident chain, screened inputs, checked citations, and a named sign-off. Same habit; a record that means something.