
Co-founder & CEO @OrdoLabs_ (YC F26). Building the AI-native OS for regulated labs
San Francisco
Joined June 2013
- Tweets57
- Following43
- Followers79
- Likes354
Benchmark: our agent fills a GLP bioanalytical study report from raw data. We scale it from 1 to 8 analytes, up to a 242-page report.
At 8 analytes: 94.5% right, 2.8% left blank, 2.0% missing a detail, 0.7% wrong.
Accuracy goes up as the report grows.
Replying to @juliarturc
happy to share the unglamorous end of it: AI in GLP labs, where every number in a report has to trace back to the raw data. dms open
2
77
Replying to @deathtouch2k
the useful version is less exciting. not skipping the trial, but turning its mountain of data into the submission faster, with every number traceable
14
Replying to @malikules
and in a regulated lab: same 0.07%, plus 400 pages proving where every number came from
2
192
Replying to @hannahhaina
lab for us. the CROs running GLP studies are the ones writing the reports and doing the QC, so they feel it first
20
Replying to @progerallworld @OrdoLabs_
the step after the analysis. turning raw instrument data into the study report takes scientists weeks, so that's where we started
10
Quick intro, since we launched this week.
I'm Sebastian, CEO of @OrdoLabs_ (YC F26), ex-quant at Handelsbanken. Regulated labs have been locked out of AI. We're building the platform that finally lets them in.
11
21
434



