Protecting Users, Engineers, and Company
Standards of Care: Responsible AI for Coaching Conversations
25,000+
Monthly conversations
Multiple (finance, conversations, pitch)
Coaching agents
Company-wide
Framework adoption
Recorded and edited by me. It demonstrates an AI system; it wasn’t made by one.
AI coaching agents handle sensitive territory daily. I designed the framework for how they should behave -- non-judgmental coaching that still redirects self-destructive thinking, protecting users, engineers, and company.
Problem Space
25K+ monthly coaching conversations entering sensitive territory, no framework for boundaries, engineers exposed to distressing logs.
What I Did
Designed Standards of Care addressing three stakeholders: users, engineers, company.
What I Built
- Standards of Care framework
- Content Moderation policy
- Distinctive conversational voice
- Hundreds of embedded boundary decisions
Adoption
All coaching agents, 25K+ monthly conversations, company-wide policy.
Outcome
- Authored Soul Machines' Content Moderation policy
- Proactive protection built before incidents
Why This Matters
Operational AI ethics -- hundreds of specific decisions, not abstract policy. Framework that actually ships.
AI EthicsContent ModerationCoachingConversation DesignPolicy