#bloodwork

Here's why nothing changes: tracking and diagnosing are different jobs, and the healthcare system only does one of them.

A fifteen-minute primary care visit exists to catch what's obviously wrong and hand out a referral. Nobody is paid to pull up your last six panels and say "your fasting glucose has climbed eight points a year for four years, let's actually deal with this."

Patient portals show one test at a time, in the format the lab sent it. Switch labs or move countries and your history shatters. ChatGPT will explain a single PDF, but it doesn't remember the last one — every panel is the first panel.

So we built BloodSight.

Upload any blood test PDF. Any lab, any country, any language. We read it, figure out what each value actually is (the same biomarker shows up as "TSH," "Thyroid Stim. Hormone," "ТТГ" depending on where it was drawn), normalize the units, and drop it onto your personal timeline.

What you get is the thing the system never assembles for you: your actual history. Every biomarker, every test, lined up. Real trends. Real deltas. The values that look fine in isolation but have been drifting for years.

A few things we cared about:

works with messy real life — different labs, languages, units, scanned printer PDFs

family accounts, so you can track your kids, partner, aging parents in one place

your data stays yours; we don't train on it, aggregate it, or sell it

PDF parsing finally got good enough. Biomarker normalization got good enough. The missing piece was someone willing to ship this for individual people instead of licensing it to hospitals for $40k a seat.

If you've ever opened a PDF of your own labs and thought I have no idea what to do with this — that's why this exists.