We're living in this crazy time where MCP lets us connect more data to AI than ever before - and honestly, it's game-changing for understanding ourselves.
But here's what's been bugging me: Garmin makes these incredible sensors that track literally everything about your body, yet their analytics feel like they're stuck in 2010. You get basic charts and generic insights that tell you nothing useful.
Like, my sleep was trash last night - but WHY? Training readiness is low - but what's the actual connection to what happened yesterday?
So I built Garmy - an open-source Python library that pulls all your Garmin data and connects it to AI through MCP. Now I can actually have conversations with my health data.
Real examples from using it:
- Going to bed after 2am drops my next day's training readiness by 15-20 points (who knew!)
- As a solo entrepreneur, I was coding late + training hard during the day - classic burnout recipe. AI spotted the pattern between my sleep quality, workout recovery, and stress levels way before I felt completely fried. Now I have actual data-driven boundaries instead of just "I'll sleep when I'm dead"
- Stress of all-night coding kills my sleep quality through HRV and REM - Claude found that connection, not me!
- Export data to my coach, now they see the full picture instead of just activities
The AI conversations are natural:
"What's messing with my sleep?"
"Should I train hard today or am I pushing too much?"
"Why has my recovery been garbage for three days?"
Instead of staring at charts trying to connect dots, I just... ask. And get real answers that actually help me not burn out.
But here's what's been bugging me: Garmin makes these incredible sensors that track literally everything about your body, yet their analytics feel like they're stuck in 2010. You get basic charts and generic insights that tell you nothing useful.
Like, my sleep was trash last night - but WHY? Training readiness is low - but what's the actual connection to what happened yesterday?
So I built Garmy - an open-source Python library that pulls all your Garmin data and connects it to AI through MCP. Now I can actually have conversations with my health data.
Real examples from using it: - Going to bed after 2am drops my next day's training readiness by 15-20 points (who knew!) - As a solo entrepreneur, I was coding late + training hard during the day - classic burnout recipe. AI spotted the pattern between my sleep quality, workout recovery, and stress levels way before I felt completely fried. Now I have actual data-driven boundaries instead of just "I'll sleep when I'm dead" - Stress of all-night coding kills my sleep quality through HRV and REM - Claude found that connection, not me! - Export data to my coach, now they see the full picture instead of just activities
The AI conversations are natural: "What's messing with my sleep?" "Should I train hard today or am I pushing too much?" "Why has my recovery been garbage for three days?"
Instead of staring at charts trying to connect dots, I just... ask. And get real answers that actually help me not burn out.
Links: GitHub: https://github.com/bes-dev/garmy
Who else wanted better analytics from their Garmin data? Curious what you'll discover!