A local archive for your Garmin Connect data.

A simple question, and what follows.

I'd read about Garmin Chat Connector, a cloud-based service that let you ask an AI about your Garmin data — a genuinely good idea, and the spark for everything that followed.

"I didn't want to hand this data to an AI company. I wanted to prepare and analyze it locally."

Could that actually work?

Simple, but working.

Three hours later there was a handful of small scripts, a dashboard, and an Excel export. It worked — faster and more easily than I'd expected.

At this point it was just a data collector. Fuel for the AI analysis that was supposed to come next.

Something didn't add up.

Comparing files from different devices and years showed a pattern the collector script had nothing to do with. My current device consistently delivered full detail. My older devices varied wildly — sometimes close to that, sometimes barely a fraction of it.

Not a bug. My older devices simply carried fewer sensors — and never sent minute-by-minute detail to begin with.

Then came the harder discovery: it wasn't only about old devices. Even my current watch loses its full resolution — Garmin quietly reduces intraday detail after several months.

Your data stays on your machine.

No extra account for GLA. No subscription. No vendor lock-in.

• No data leaves your machine unless you export it yourself
• The archive itself is plain, machine-readable data — not locked to GLA
• Encrypted dashboards and an encrypted mirror container, for when you do want to move the archive between machines