Lumose

Between endocrinology appointments, most people with diabetes are on their own — staring at glucose graphs, guessing at patterns, wondering what they're missing. That gap is where Lumose was born.

Lumose is an open-source diabetes management platform that helps you understand your data between appointments — and at 3am, when no one else is looking.

It connects your CGM and insulin pump to an AI analysis engine that surfaces your patterns, flags what looks off, and helps you walk into your next appointment with real questions instead of guesses. Daily briefs, meal analysis, trend recognition, a conversational AI chat grounded in clinical diabetes knowledge, and a configurable alerting system that surfaces issues worth your attention.
 

Predictive-trend alerts, configurable thresholds, caregiver escalation, and multi-channel delivery, so the right person can see the right thing at the right time.

It helped its creator make sense of his own data while waiting months to be seen — and it's built so the next person doesn't have to fly blind in that gap.

The platform is self-hosted, privacy-first, and BYOAI: you bring your own AI provider, you own your data, and you're never locked into a single vendor. Run it fully local with Ollama, or connect Claude, OpenAI, or any compatible endpoint.

Lumose integrates directly with Dexcom and Tandem, and connects to the wider diabetes ecosystem through Nightscout — pulling in data from Loop, AAPS, and Trio — with support for Medtronic and Glooko, plus a plugin SDK for community-built device drivers. Companion Android and Wear OS apps put your data on your wrist.

Funds raised here support maintainer compensation, project infrastructure, and developer platform licensing (GitHub, Apple, Google) as we work toward official app-store releases alongside the open-source sideloaded builds.
 

Lumose is not a medical device and doesn't replace professional medical care or your own judgment — it helps you see your data more clearly and have better conversations with your care team.

Built by a patient who was tired of feeling like he was on his own. 
Open-source, so the next person doesn't have to build it alone. Stay safe. 💙