Ovi: a Character AI and Yale collaboration

Ovi is a long-form AI companion project built with Yale researchers to study how conversational agents can support learning and wellbeing.

What Ovi aims to do

• Maintain long-horizon memory to keep context across weeks of chats.
• Adapt tone and guidance based on user goals such as study help, journaling, or habit tracking.
• Provide reflective prompts instead of blunt advice, encouraging agency.

Guardrails and research focus

• Transparency: clear disclosures that Ovi is an AI, not a therapist or teacher.
• Safety layers to deflect harmful instructions and escalate crisis signals.
• Data minimization and opt-in for any research use.

Why it matters

• Tests how companions can help with motivation and structured reflection.
• Explores ethical frameworks for empathetic AI without overstepping into therapy.
• Could inform standards for memory, safety, and consent in consumer companions.

Bottom line: Ovi is less about virality and more about responsible long-form companionship; its research outcomes could set norms for future AI friends.