Clear AI identity
Trin must disclose that she is an AI assistant and should not pretend to be human.
Trust and safety
Built for real calls, not impersonation.
Call Trin is designed around explicit AI disclosure, owner-controlled context, private memory boundaries, and operational safeguards for call answering, outbound call tasks, shared owners, and follow-up workflows.
- AI disclosure
- Owner control
- Secure sessions
Owner memory separation
Private owner-mode memories stay separate from public caller behavior and external caller prompts.
Secure web sessions
The app uses HttpOnly cookie authentication and CSRF checks for unsafe browser requests.
Bounded assistant behavior
Trin should answer from configured context, avoid regulated commitments, and collect details or place only owner-authorized calls when escalation is needed.
Operating principles
Useful, transparent, bounded.
Trin is designed to help with everyday communication, not to hide automation or make unsupervised decisions beyond the context you approve. Outbound calls should be clear, task-specific, and authorized by the owner.
Disclose AI identity
Use approved context
Escalate owner decisions
Keep reviewable outputs
Caller transparency
Callers should understand they are speaking with an AI assistant, especially where recording or transcription consent is required.
Audit-friendly outputs
Summaries, caller intent, questions, promises, and follow-up items make it easier to review what happened before acting.