Safety and evidence
What are the limits of AI HYROX coaching?
AI can organise information, explain training concepts and draft sessions. It cannot diagnose injury, feel movement quality, guarantee that a limiter is correct or prove that one workout improved performance.
Use AI for structure, not authority
A useful system treats AI output as a proposal that must pass deterministic rules, athlete context and safety constraints. Confidence, polished language and personalisation are not evidence that a prescription is correct.
AI cannot safely replace
- A clinician assessing pain, illness or injury.
- A coach observing technique and behaviour in real time.
- Verified race rules, division standards or current event instructions.
- A progression model that checks time, distance, load, rest and recovery.
- The athlete's judgement when symptoms or conditions change.
How LimiterIQ narrows the risk
LimiterIQ uses deterministic plan rules and explicit safety gates around AI-assisted coaching. It records what is observed and avoids claiming improvement from a single completion. That reduces risk but does not remove the limitations above.
Strava boundary
LimiterIQ is seeking written policy clarification about operating AI features alongside technically isolated Strava data. Broad Strava-led promotion remains paused. Policy clearance has not been claimed.
Method and ownership
Written by LimiterIQ and reviewed 27 July 2026. This page describes LimiterIQ's intended safeguards and limitations. It is not an independent audit, medical advice or proof that every software path is error-free.