AI Transparency

When you chat with an AI coach in Mentoré, here's what happens:

1. Context Gathering: We retrieve relevant context from your recent journal entries, tracking data, and conversation history
2. Prompt Construction: Your message and context are combined with the coach's personality and instructions
3. AI Processing: The prompt is sent to an AI model which generates a response
4. Response Delivery: The response is sent back to you and saved in your conversation history

Mentoré partners with leading AI providers to power coaching conversations:

• OpenAI (GPT-4): Primary AI model for coaching conversations
• Anthropic (Claude): Alternative model for certain coaching styles

We select AI providers based on their commitment to data privacy and their enterprise-grade security practices. All providers have:

• SOC 2 Type II certification
• Data Processing Agreements (DPAs) with us
• Zero data retention policies for API calls
• No training on customer data

When you send a message to an AI coach, the AI may receive:

• Your current message
• Recent messages in the current conversation
• Relevant excerpts from recent journal entries (typically last 30 days)
• Recent tracking data (mood, sleep, habits) if relevant
• Coach-specific personality and instructions

What AI Does NOT See

• Your email address or account credentials
• Your payment information
• Other users' data
• Entries from years ago (unless specifically relevant)
• Data you've deleted

Specifically:

• OpenAI: We use their API with data usage opt-out. OpenAI does not train on API data.
• Anthropic: Enterprise agreement prohibits any training on customer data.

It's important to understand what AI coaches can and cannot do:

AI CAN:

• Provide thoughtful reflections based on your context
• Identify patterns in your tracking data
• Offer different perspectives on situations
• Remember what you've shared in recent conversations
• Adapt communication style to different coach personalities

AI CANNOT:

• Provide medical, legal, or financial advice
• Diagnose mental health conditions
• Replace professional therapy or counseling
• Guarantee accuracy of all statements
• Access real-time information (news, weather, etc.)
• Remember information between separate sessions perfectly

You have control over how AI interacts with your data:

• Delete Conversations: You can delete any conversation, removing it from your history and future context
• Delete Journal Entries: Deleted entries are removed from AI context
• Choose Your Coach: Select coaches with different personalities and communication styles
• Export Your Data: Download all your data at any time
• Delete Your Account: All data is permanently deleted

We implement safety measures to ensure AI responses are appropriate:

• AI providers have built-in content safety filters
• Coach prompts include guidelines for appropriate responses
• Crisis detection to suggest professional resources when needed
• Harmful content monitoring and flagging

If you encounter an inappropriate AI response, please report it to us.

We commit to the following principles as AI technology evolves:

1. Transparency: We will always be clear about how AI works and what data it uses
2. Privacy First: Your data will never be used to train public AI models
3. User Control: You will always have control over your data and how AI uses it
4. Continuous Improvement: We will update our practices as AI safety research advances
5. Honest Limitations: We will be honest about what AI can and cannot do