What's the most painful part of building a health app?
Asking because we've shipped 220+ health tech projects at Momentum and the answer is almost always the same: wearable integrations.
Not the AI. Not the UX. Not the business model. The data layer.
Garmin, Whoop, Apple Health, Oura - every provider has a different auth flow, different data format, different rate limits, different webhook structure. Garmin sends XML. Whoop sends JSON but paginates differently than Oura. Apple Health needs on-device processing. Samsung has its own SDK.
A team of 3 engineers can spend 3-4 months just getting data flowing from 3-4 providers. Then another month normalizing it so you can actually compare a Garmin sleep session with a Whoop one. That's 4-5 months before you've written a single line of actual product code.
And then you realize you still need to build health scoring. What does "good recovery" mean for your users? How do you calculate it consistently across 5 different devices? What's the HRV baseline for a 60-year-old versus a 25-year-old? Another few months of research and engineering.
That's why we built Open Wearables and launched it today on Product Hunt. Open-source platform: one API for multiple wearable providers, open health scoring algorithms you can actually read and modify, and an AI reasoning engine via MCP server. MIT licensed, self-hosted, $0 per user.
The goal was simple: the boring infrastructure should be free and open so teams can spend their time on what actually makes their product different.
Curious what others have run into. What killed the most time on your health project? And if this resonates, would love your support today.
Replies
I’ve been through this , and I completely agree, the data layer eats your timeline alive. for me , authentication flows alone were a nightmare.
@deangelo_hinkle do you also handle historical backfills cleanly?
@deangelo_hinkle @henry_lindsey honestly, I underestimated how messy wearable data is until I built something myself. I thought APIs would be the easy part, but consistency killed me.
@deangelo_hinkle For me, the hardest part was explaining to stakeholders why “just connecting devices” takes months. It sounds simple until you hit rate limits, missing fields, and weird timestamps.
@deangelo_hinkle @shawn_idrees how flexible is it if I want to plug in my own models later?