LessonLens analyses what happens in your classroom without ever storing — or showing — who said it. Here's how the safeguards work, end to end.
Student names are stripped from every transcript before it touches our database. Both intake paths — pasted transcripts and AssemblyAI webhooks — call the same redaction step, and it's fail-closed: if redaction errors, we discard the transcript rather than store an unredacted copy.
The recording detail page has no transcript view. The redacted copy exists for AI analysis only, so even an authorised mentor or HoD never sees the words your students said — just the AITSL evidence drawn from them.
Mentors only see lessons that have been explicitly shared with them, and only while an active mentorship exists. Heads of department see aggregated activity across their team — never individual transcripts or comments.
Our database, storage and edge functions run in Australian regions. Audio files transit AssemblyAI for transcription, then transcripts are redacted on our infrastructure before being stored.
We're transparent about what's still being hardened before wider rollout.
Have a security question that isn't answered here? Check the FAQ or reach out via your pilot contact.