AI & Data Policy

1. Overview

This AI & Data Policy explains how WellVoye uses artificial intelligence to generate wellness coaching, what data is shared with AI systems, how we protect your privacy during AI processing, and what safety measures are built into our AI pipeline. This policy supplements our Privacy Policy and Terms of Service.

2. AI Providers

WellVoye uses a multi-provider AI architecture to deliver the best possible coaching experience:

2.1 Anthropic (Claude)

Used for nuanced wellness coaching conversations, trend analysis, and personalized recommendations. Claude excels at thoughtful, safety-conscious responses.

2.2 OpenAI (GPT)

Used for complementary coaching capabilities, alternative perspectives, and specific wellness analysis tasks.

2.3 Apple On-Device Intelligence

Used for local processing tasks that do not require cloud connectivity. On-device models enable real-time insights without sending any data to external servers.

2.4 Automatic Fallbacks

If a cloud AI provider is unavailable, WellVoye automatically falls back to alternative providers or on-device deterministic coaching algorithms. This ensures you always receive guidance, even without an internet connection.

3. What Data AI Accesses

When generating personalized coaching, the AI may receive:

• Aggregated wellness metrics — step counts, sleep scores, hydration levels, and other pillar data in summary form
• Trend data — patterns over days or weeks (e.g., "sleep duration decreased 15% this week")
• Voyage progress — your current stage in active wellness programs
• Conversation context — previous messages within a coaching session

4. PII Sanitization Process

Before any data is sent to an external AI provider, WellVoye applies a multi-layered sanitization process:

• Email detection & removal — any text matching email patterns is stripped
• Phone number redaction — phone numbers in all international formats are detected and removed
• SSN/ID filtering — social security numbers and government ID patterns are blocked
• Credit card scrubbing — credit card number patterns are detected and removed
• Name anonymization — personal names are replaced with generic references
• Location abstraction — specific locations are generalized (e.g., city-level weather context only)

This sanitization happens entirely on your device before any data leaves the app. The AI providers never receive unsanitized data.

5. AI Safety Rules

WellVoye implements a 6-rule AI safety filter that governs all AI-generated responses. Every output from the AI is evaluated against these rules before being shown to you:

6. Data Provenance & Tracking

WellVoye maintains data provenance tracking, which means we record:

• Which AI provider generated a given response
• The model version used at the time of generation
• Whether the response was cloud-generated or on-device

This allows us to audit AI outputs for quality and safety, and to quickly address any issues with specific providers or model versions.

7. User Consent & Control

You have full control over how AI is used in your WellVoye experience:

• Opt-in required — AI coaching features require your explicit consent before activation
• Disable AI anytime — you can turn off cloud AI processing and rely solely on deterministic coaching and on-device intelligence
• Conversation history — you can clear your AI coaching conversation history at any time
• Provider preference — where available, you can choose your preferred AI provider

8. Data Retention by AI Providers

WellVoye's agreements with AI providers include the following protections:

• AI providers do not use your data to train or improve their models
• Conversation data sent to providers is processed in real-time and not retained beyond the session
• No user profiles are built by AI providers from WellVoye interactions

9. Deterministic Fallback

When AI services are unavailable or when you choose to disable AI, WellVoye provides deterministic coaching based on:

• Pre-programmed wellness algorithms and thresholds
• Evidence-based recommendation engines
• Static coaching content curated by wellness professionals

This ensures you always receive valuable guidance regardless of AI availability.

10. Continuous Improvement

We are committed to continuously improving our AI safety and privacy practices. This includes:

• Regular audits of AI outputs for safety and accuracy
• Updating sanitization rules as new PII patterns are identified
• Evaluating new AI providers against our privacy and safety standards
• Incorporating user feedback into our AI safety framework

11. Changes to This Policy

We may update this AI & Data Policy as our AI capabilities evolve. Material changes will be communicated through the app. The "Last updated" date at the top indicates the most recent revision.

12. Contact Us

For questions about our AI practices or this policy, please contact us:

• Email: support@wellvoye.com
• Developer: Madhusudan Bansal / InSynapse