AI, honestly
If you run a program that serves people, you should interrogate any software that puts AI near their records. Here are the questions we get in every first meeting — and our real answers.
HIPAA-aware by design. All data is treated as sensitive, behind login and role-based access — a per-capability permission matrix your admins control.
No. AI drafts interview scores and case summaries. Your staff review them and make every decision. There is no automated action in Clair that changes a participant's status.
Everything is reviewable. Staff can read the source transcript, edit any output, and regenerate it. Nothing AI writes is hidden from the people accountable for it.
Deterministic fallbacks keep critical outputs flowing — eligibility summaries generate even when no model is reachable. Your operations never wait on a model.
You write the instructions. Tone, focus, length, and behavior for every AI interviewer and every summary — set by your admins, changeable any time.
Fully bilingual. The interface, emails, SMS, and every AI conversation switch per user, per participant, per session.
Clair runs on its own servers. Your program's data is not pooled with, processed alongside, or routed through other companies' AI products.
The AI in Clair works only inside Clair. Your transcripts, notes, and summaries are never fed into external AI assistants, plugins, or aggregators.
Model-training opt-out is enforced at the Clair platform level for every customer, every record, by default. You don't have to find a checkbox. There isn't one to find, because it's already off.
These aren't settings. They're how Clair is built.
Nothing is off the table in a first meeting. That's the point of one.
Book a demo