The pharma founder who needed investors to believe the science
Predicting drug bioequivalence from messy clinical PDFs, starting from a HIPAA foundation.
GenRx had a bold idea: predict drug bioequivalence from messy clinical PDFs. But an idea isn't an architecture, and investors fund what they can see.
They needed to walk into a funding round with a secure, working AI architecture, not a slide.
We started with the foundation, a HIPAA-aligned environment on AWS, before a single prediction ran.
An engine that reads unstructured clinical PDFs (NLP), predicts the key pharmacokinetic metrics (PyTorch), investor-ready.
5.0 / 5.0 on Clutch, quality, schedule, cost.
In regulated domains, the model is the easy part. The architecture that makes it safe and credible is the product.
The car factory still picking parts on paper
A high-volume assembly line that still picked parts by hand off paper lists.
The security agent that worked in the demo, and broke in the real world
A macOS compliance agent that ran fine interactively, but wasn't something they could ship.
One API over many gateways, without pulling the whole app into PCI scope
A US gateway provider drowning in bespoke processor integrations.
From reactive to predictive, on systems actually live in production
A US B2B services operation running on disconnected systems with no single source of truth.
When matchmaking lived in a few people's heads
A Singapore innovation ecosystem matched startups to investors by hand.
Scaling remote care without scaling the overhead
TELEMED2U needed to scale remote care, but consultations were inefficient and costs were climbing.
One records platform across a growing clinic network
A clinic network running on siloed records that couldn't keep up as it grew.