Safety Detection

Veterinary visits occasionally go sideways — an agitated client, threats, or refusal of care that might later become a dispute. Noctune watches for these and quietly preserves what happened so you’re protected. This costs you nothing and is on by default for every account.

What we flag

During analysis of a transcript, Noctune detects:

• Hostile language directed at you, your staff, or the clinic
• Threats — physical, legal, or financial
• Explicit refusal of care — so there’s a clear record of what was offered vs. declined
• Sentiment trends that escalate across a conversation

Each flagged encounter gets a small indicator on the encounter card and on the patient’s profile.

What happens automatically

When a visit is flagged:

1. The audio is preserved indefinitely — pulled out of the normal retention/archival schedule, no extra cost to you.
2. A snapshot of the transcript and SOAP note is locked — edits after the flag fires are tracked as new versions, originals stay intact.
3. A flag appears on the patient’s profile — future-you (or another vet on the team) sees it before the next visit. See Patients.
4. The encounter is searchable in the Messages inbox under the “Flagged” filter so you can find it quickly.

None of this shares your data with anyone outside your practice. We don’t notify authorities, insurers, or third parties — it’s entirely a protective record for you.

Opting out

You can mute the indicator on any individual encounter if it was a false positive (Encounter → three-dot menu → Unflag). The underlying audio preservation and lock stay in place — unflagging only hides the indicator. You can’t opt out account-wide, because the protection exists for you.

Reviewing flagged encounters

The Messages inbox has a Flagged filter that surfaces every encounter with an active safety flag, sorted newest first. The patient profile also shows a historical hostile-encounter count so you can read the room before seeing a repeat client.

What we don’t do

• No recording without your knowledge. Safety detection runs on audio you already chose to record. It doesn’t turn on the mic, extend recording time, or transmit extra data.
• No sharing. Flagged data stays in your practice unless you explicitly export or forward it.
• No AI-generated testimony. Flags are a signal, not a legal determination. What gets preserved is your own audio and transcript — the source of truth.