The R&D thesis

Philip and I met at an On Deck hacker house in fall 2020. We were both repeat founders. We were both drawn to the same kind of problem: messy source material, high-stakes decisions, and no good software in the middle.

Pharma and scientific R&D looked like the obvious first target. The industry is one of the most acquisitive in the world. Labs generate enormous volumes of unstructured data. The workflow tools available to researchers were genuinely bad. I had spent enough time around energy research at Princeton and 24M to know what broken research infrastructure looks like. This looked like a clear opening.

We iterated through several product shapes. Electronic lab notebooks. Project management. Data integration. Alliance-management software. Each version taught us something about what teams actually needed. The common thread across all of them was the same: large volumes of messy information, many stakeholders, and work that falls apart if the system is not trustworthy.

None of that iteration was wasted. It is where we built file categorization, natural-language search, audit logs, role-based access, and workflow features that let multiple teams work off the same source material without losing the thread. Every one of those capabilities mattered later.

The business logic, though, was wrong. Smaller biotech companies did not have enough budget for R&D software. Larger ones came with 12-month enterprise cycles, complex integrations, and spending that skewed far more toward clinical development than research operations. The product was real. The market was not.