Provider Resource Hub: App & AI
Safety guardrails
Designed with safety in mind
Both coach and the iOS app reflect privacy-by-design practices, including built-in safeguards and routine reviews that work together to help protect client information.
- While chats are not monitored in real time, we use automated evaluation systems and have a team of licensed clinicians conducting routine reviews that help monitor and uphold quality and safety standards.
- Our safeguards within the app block the conversation and alert you in your portal if the system detects a potential safety risk. You can then evaluate the situation and determine whether it’s appropriate to allow the client to regain access to coach.
You’re informed and in control
- You can view client conversations with coach by default to monitor discussions. Clients can enable private mode for conversations they’re not ready to share with you yet. However, if safety concerns are identified in private or public mode, coach is disabled and you’re sent an alert in your portal for clinical assessment.
- You can disable coach for a client if it’s not appropriate for their care.
- Data is not shared externally. We do not use coach conversation data to train or improve external AI models.
- In-session and between-session data (such as coach chats and therapy toolkit use) is stored during active care and deleted 60 days after care ends, enhancing continuity while maintaining appropriate data retention. Medical records remain separate.
When coach may not be appropriate
Clinical contraindications
- Active psychosis or severe mental health crisis
- Clients with poor insight or reality testing
- Severe personality disorders with unstable attachments
- Clients who struggle with boundaries in therapeutic relationships
Provider discretion factors
- Treatment modality incompatibility (approaches that require strict provider-only contact)
- Client preference for traditional therapy only
- Cultural or religious considerations
- Complex trauma requiring careful therapeutic pacing