AI Usage Policy
Last Updated: June 26, 2026
This AI Usage Policy describes how artificial intelligence is used within the GetFit mobile application operated by KeyZen.
1. AI Features in GetFit
GetFit incorporates the following AI-powered features:
1.1 AI Trainer Chatbot (Kyro Pro+ Feature)
- A conversational AI fitness and nutrition coach available to Kyro Pro+ subscribers
- Provides personalized workout advice, nutrition guidance, motivation, and answers to fitness-related questions
- Adapts its communication style and recommendations based on your profile, conversation history, and stored memories
1.2 AI Diet Plan Generator (Kyro Pro Feature)
- Generates personalized diet plans based on your health profile, fitness goals, dietary preferences, allergies, health conditions, cuisine preferences, cooking time, and budget
- Calculates appropriate calorie targets and macro distributions
1.3 AI Food Recognition
- Identifies food items from photos captured by your device camera
- Estimates nutritional content (calories, protein, carbs, fat) of identified foods
- Supports identification of multiple food items in a single image
1.4 AI Video Form Analysis
- Analyzes exercise form from user-submitted videos
- Detects exercise type, counts repetitions, and provides a form score
- Offers feedback on form improvement
1.5 AI Memory System
- The AI Trainer extracts and stores factual information from your conversations (e.g., "has a knee injury," "prefers morning workouts," "is vegetarian")
- Memories are categorized by type: injury, goal, preference, body stats, experience, limitation, achievement, routine, nutrition, progress
- Memories have a hierarchical system:
- Level 1 (Core Identity): Very slow decay — allergies, chronic injuries, fundamental preferences
- Level 2 (Long-term Evolving): Medium decay — weight, routines, progress
- Level 3 (Short-term Context): Fast decay (24-72 hours) — temporary states like soreness, travel
- Level 4 (Session): Destroyed when the chat session ends
1.6 Adaptive Coaching
- The AI maintains a "User State" that tracks inferred metrics such as energy, recovery, fatigue, motivation, adherence, injury risk, burnout risk, and plateau risk
- These metrics are used to adjust AI coaching recommendations (e.g., suggesting deload periods when fatigue is high, simplifying plans when adherence is low)
- The AI may predict behavioral patterns such as likely plan abandonment, motivation drops, or readiness for progression
2. AI Technology
- GetFit uses a custom self-hosted AI model for all AI features
- The AI service processes your data in real-time and does not share your data with third-party AI providers (e.g., OpenAI, Google, Anthropic)
- Food recognition uses computer vision models to identify food items from images
- The AI model is hosted on infrastructure controlled by KeyZen
3. Data Used by AI
The AI features use the following data to provide personalized responses:
| Data Type |
How Used |
| Your profile (name, age, gender, weight, height, goal, diet preference) |
Personalizing recommendations |
| Your subscription tier |
Determining available AI features |
| Chat messages you send |
Generating relevant responses |
| AI-extracted memories |
Providing context-aware coaching |
| Food images (camera captures) |
Identifying food items |
| Exercise videos (user-submitted) |
Analyzing exercise form |
| User state signals (workout logs, meal logs, feedback) |
Adapting coaching intensity and tone |
4. Your Control Over AI Data
You have full control over your AI data:
- View memories: See all facts the AI has stored about you
- Delete individual memories: Remove specific facts you don't want the AI to remember
- Confirm memories: Verify AI-extracted facts for higher accuracy
- Reset all memories: Clear your entire AI memory profile
- Export memories: Download all AI memories associated with your account
- Delete chat sessions: Remove conversation history
- Delete account: Removes all AI data permanently
5. AI Limitations and Disclaimers
⚠️ IMPORTANT: The AI is NOT a medical professional, licensed dietitian, certified personal trainer, or healthcare provider.
5.1 No Medical Advice
- AI responses are for informational and educational purposes only
- AI-generated diet plans, calorie recommendations, and fitness advice are estimates based on general formulas and may not be appropriate for your specific medical conditions
- Always consult a qualified healthcare professional before making significant changes to your diet or exercise routine
5.2 Accuracy Limitations
- AI food recognition may misidentify foods or provide inaccurate nutritional estimates
- AI diet plans may not account for all medical conditions, medication interactions, or individual nutritional needs
- AI exercise form analysis may miss critical form errors that could lead to injury
- AI memory extraction may misinterpret or misremember facts from conversations
5.3 Not a Substitute for Professional Services
The AI Trainer is not a substitute for:
- A licensed physician or medical professional
- A registered dietitian or nutritionist
- A certified personal trainer or exercise physiologist
- A licensed mental health professional
5.4 User Responsibility
You are solely responsible for:
- Verifying the accuracy of AI-generated information before acting on it
- Disclosing relevant health conditions and allergies when interacting with the AI
- Discontinuing any AI-recommended plan that causes adverse effects
- Seeking professional advice for serious health, nutrition, or fitness concerns
6. AI Training and Data Usage
- Your conversations and data may be used to improve the AI model's accuracy for your own personalized experience
- We do not use your personal data to train AI models that are shared with or used by other users
- AI model improvements are focused on system-level quality, not individual user data exposure
- You can opt out of AI data retention by deleting your AI memories and chat history
7. AI Safety
We implement the following safety measures:
- Memory validation: AI-extracted memories are validated before storage to prevent erroneous facts
- Memory quarantine: Uncertain or emotionally-driven facts are quarantined and require verification
- Contradiction detection: The AI resolves conflicting memories to maintain accuracy
- Truth decay: Memory confidence scores decay over time if not reinforced, ensuring outdated information fades
- Rate limiting: AI endpoints are protected against abuse
- Content filtering: The AI is designed to refuse requests for harmful, abusive, or medically dangerous advice
8. Changes to AI Features
We may update, modify, or discontinue AI features at any time. We will notify users of material changes through the App.
9. Contact
For questions about our AI features or this policy:
This AI Usage Policy is effective as of June 26, 2026.