These notes explain the public problem behind the foundation’s focus. They are not announcements of completed foundation studies; they describe why clinical trials in a dish matter for AI-era drug discovery.
June 4, 2026
What a “trial in a dish” should prove
A dish-scale experiment should do more than show that a biological model is complex. It should answer a decision question: whether a candidate is worth advancing, whether a mechanism deserves more work, or whether a weak signal should stop consuming resources.
This is why benchmarks matter. Without a stated context of use, a sophisticated organoid or chip assay can become a demonstration that does not change the evidence chain. The standard should be stricter: what is measured, how reproducible it is, and what decision it informs.
Research records. Open evaluation notes make preclinical evidence easier to compare and harder to overstate.
May 29, 2026
The candidate bottleneck after AI generation
AI systems can expand the number of plausible therapeutic candidates. That is useful, but it moves pressure downstream. More candidates mean more decisions about which hypotheses deserve biological validation, animal work, manufacturing attention, and eventually clinical enrollment.
The foundation’s position is that candidate generation and candidate triage should be designed together. If the evidence layer remains thin, the field risks producing more candidates without improving the quality of candidates that reach patients.
