trainedby.ai connects your wearables to ChatGPT, so your AI coach knows your real training load, sleep patterns, and recovery without you typing a word.
The problem with AI fitness coaching today
People are turning to ChatGPT to plan their marathon training, strength programs, or HYROX prep. It makes sense, because AI is available 24/7, extremely patient, and quite knowledgeable.
But somehow it doesn't quite work.
And that is because you have to tell it everything. Every. Single. Time.
Every conversation starts from zero. You have to describe in detail your current level, recent workouts, schedule, and goals. You paste in screenshots from Garmin. You try to remember what you did last week, how you slept, that your resting HR has been elevated, or that your HRV has been trending down for weeks.
The advice is based on what you told it, not what actually happened.
And what you tell it is incomplete, subjective, and often wrong.
What if your AI coach could see everything?
We connect your wearables (Garmin, Oura, Whoop, Apple Health, and 50+ other platforms) directly to your AI assistant.
When you open ChatGPT and ask about your training, it already knows:
No more context dumps. No more subjective self-reporting.
How it works
We use smart compression to turn weeks of wearable data into something an AI can understand and analyze. Instead of dumping raw heart rate readings into a chat window, we surface the trends and patterns that matter—elevated resting HR, declining HRV, increasing training load, poor sleep consistency.
The data rolls up from days to weeks to months to years, so you can spot trends.
And because it's integrated into ChatGPT, you just start chatting.
Beyond wearable data, you can add notes like workout goals, feedback, and other personal information such as availability and injuries. This way, the AI gets that bit of subjective context it needs to give you the best advice.
Who this is for
If you're already wearing a watch or ring, and using AI to help plan your training, whether that's for a marathon, strength program, or general fitness, trainedby.ai makes that experience dramatically better.
Get started
The product is free to use.
Connect your wearables, add the integration to your AI assistant, and start chatting.
get_onboardingGet all profile-building questions ordered by importance. Returns questions sorted by weight, where higher weights indicate questions with fewer relevant notes in the timeline. This identifies areas needing more coverage. Always pre-fills answers from timeline data when available, but ask user to verify.
get_timelineRetrieve a timeline of AI-generated summaries for goals, notes, and health data. For period granularities (day, week, month, year), retrieves looking back from today. For item granularity, retrieves individual summaries for a specific date. Each summary includes source_count: the number of individual items (activities, goals, notes, etc.) that were aggregated into that summary. Never returns future periods.
granularitystringTime granularity - 'item', 'day', 'week', 'month', or 'year'
datestringDate in YYYY-MM-DD format (required when granularity is 'item')
nullrange_sizeintegerNumber of periods to retrieve (required for day/week/month/year, max: 7 days, 5 weeks, 12 months, 10 years)
nullsave_noteSave a new note. Notes can be goals, workout feedback, reflections, or general observations. These notes are used to build your profile and provide personalized coaching.
contentstringThe note text content
note_typestringType of note (goal, workout_feedback, reflection, general)
datestringOptional date associated with the note in ISO format (YYYY-MM-DD or ISO 8601 timestamp). Defaults to current time if not provided.
nullsearch_timelineSearch timeline summaries by semantic similarity using vector embeddings. Searches across all summaries (health metrics and notes) at the specified granularity level. Returns results ordered by relevance (similarity score).
querystringText to search for across timeline summaries
granularitystringGranularity level to search in ('item', 'day', 'week', 'month', or 'all')
alllimitintegerMaximum number of results to return (default: 20, max: 100)
20share_user_feedbackShare feedback about this MCP with the developer. Use this to report bugs, request features, ask questions, or share what's working well. Your feedback directly helps improve the product.
feedbackstringThe feedback to share
categorystringThe category of feedback (e.g., bug, feature, question, other)
othercontextstringAdditional context about where or how the issue occurred
nullurgencystringThe urgency of the feedback (e.g., low, medium, high)
lowwhoamiVerify user authentication and check data access permissions.
Returns: user email, profile ID, and whether the user has access to their data.