The short version
GRITFIT Coach responses are generated by AI. Coach is not a human coach, doctor or clinician. Its responses can be incomplete, inaccurate, inconsistent or unsuitable for you.
GRITFIT is intended for general fitness and wellness coaching. It is not intended to diagnose, treat, cure or prevent disease, provide rehabilitation, or handle emergencies. Stop exercising and seek appropriate professional or emergency help if symptoms are concerning.
How Coach works
A useful summary of the architecture is: the app asks; GRIT remembers permitted context; ATLAS reasons; an external language-model service generates language; safety gates judge the result; GRITFIT stores the relevant training outcome; and the next workout can evolve.
In practice:
1. the app sends your request to the GRITFIT backend; 2. deterministic orchestration selects a bounded set of relevant account, profile, workout and conversation context; 3. ATLAS applies training, recovery and progression logic and prepares structured instructions; 4. a protected external language-model provider generates candidate wording; 5. GRITFIT safety validators and workout-structure checks assess the result; 6. a validated response is returned, or GRITFIT uses a safer fallback or declines the request; and 7. permitted conversation and training records can inform later sessions.
These controls reduce risk but cannot guarantee correctness or safety. GRITFIT may change models or provider infrastructure while maintaining the disclosure, permission and safety boundaries described here.
Your permission and control
Before the first meaningful Coach use, the app explains that Coach is AI-generated, can be wrong, and is not a human coach or clinician. You then choose Enable Coach or Continue without Coach.
If enabled, you give GRITFIT permission to send your Coach message and only the profile, training and separately allowed health context needed for the request to its contracted external language-model provider. You can review or withdraw this permission later in Settings. Withdrawal stops new provider requests; non-AI core training remains available.
What is and is not sent
The provider request may include your message, relevant recent Coach context, fitness goals, training state, workout history and separately permitted health/recovery context needed for the request. GRITFIT removes direct account credentials and filters prohibited keys and location structures before provider processing.
GRITFIT does not send passwords, authentication tokens, Hidden Folder media, private progress media, failed-unlock security photos, or precise GPS route points for general Coach inference. The provider receives bounded context, not unrestricted access to GRITFIT databases.
The identity, contract terms, provider-side logging, training restrictions, processing locations and transfer safeguards for production language-model infrastructure are maintained through GRITFIT's processor governance. Contact dpo@gritfit-ai.com for current information relevant to a privacy request; support@gritfit-ai.com is the backup route.
Health and safety limits
Coach can discuss general exercise, recovery and training context, including information you choose to provide about pain, fatigue, injuries or limitations. That does not make the response a diagnosis or clinical recommendation. Do not use Coach to decide whether serious symptoms are safe, to replace a qualified professional, or to contact emergency services.
Safety gates can block or redirect some risky requests, but a disclaimer or validator does not make every output safe. Check exercises, loads and plans against your circumstances, equipment and professional advice.
Human involvement and feedback
Routine Coach replies are automated. Authorised GRITFIT personnel may access limited records when reasonably necessary for support, safety investigation, abuse prevention, security, legal compliance or improving a reported defect. Access is not unlimited and private media is not made part of Coach context.
You can report an unsafe or incorrect response through in-app feedback or support. Do not submit an emergency through Coach or support.