Protect your license. Automatically.

You're still the supervisor, and Clerie never signs anything for you. Risk detection is tuned deliberately sensitive, especially on self-harm: we'd rather flag a session you didn't need to read than stay quiet on the one you did. A flagged note jumps to the top of your queue. An unflagged note still gets a summary, still sits in the same queue, and still waits for your signature. The alert changes what you read first. It doesn't change who's responsible.

That's the exact problem this tool exists for. Clerie reads every supervisee note before you do and writes the summary: what they did well, what needs work, plus drafted feedback you can edit and send. Risk alerts pull the sessions involving harm, suicide, or a spiking assessment score to the top of the queue. So your hour of review goes to the three notes that actually need a supervisor, not to hunting for them across ten caseloads. You still see everything. You just stop reading blind.

No practical limit. Every note gets read and summarized whether your team wrote forty this week or four hundred, and we don't meter the AI per note or per seat. Nobody has ever run out. The same goes for risk alerts: they run on every session, not a sample, because a sampled safety net isn't one. If your clinic is big enough to worry about volume, bring a real week of sessions to the demo and watch what a heavy queue looks like.

Automatically. Every session you approve ticks the supervisee's tracker, so the count reads 53 of 2,000 on the dashboard instead of living in a paper log they ask you to double-check. Clerie also keeps the caseload demographics your board asks for, tracked in the background as sessions happen. When a supervisee applies for licensure, the record is already built: hours, session types, and your sign-offs in one place. Nothing to reconstruct at the end of two years.