5 questions from a non-exhaustive list that I'm thinking about right now

[30052025]

1

As a R & D focused venture-backed venture, how do you balance development (exploit) vs research (explore) ? Consolidating gains at the frontier and cooking them to production paved paths towards GTM and is essential to survival in the short-term as well as to dominate your blue ocean. But constantly nosing towards the blue sky is how you stay relevant in the long-term (length contingent on AI timelines), and more educated shots on goal increases your surface area for luck in the present. Cutting-edge research is also is also a powerful ambition pull towards aspiring to be world-class, elevating your team’s taste of what counts as good enough. At many points though, you will have to choose.

R / D.

2

Reinsurers are mammoths who provide insurance to insurers, i.e. they wield the actuarial sciences to provide protection against risk. What predictive algorithms are the world’s most innovative reinsurers deploying to better understand and stratify the risk of cancers in their cohorts? Given cancer is the cause of more than half of all healthcare claims above median stop-loss coverage, what would be a good estimate for the monetary value generated by a highly predictive (AUC 0.99) multi-cancer early detection risk algorithm?

3

The olfactory code is infamous for being combinatorially intractable. It is very hard to predict scents from molecules, especially if those scents are complex and shifting patterns of thousands of molecules — like human breath. Yet, biologically organisms are able to consistently identify odors. What can data from machines like GC-MS and e-noses, limited as they may be, tell us about how the sublime adaptability of bio percepts? Does noisy and variable GC-MS PPM data shed any light at all to the extremely sophistication and precision of the canine noise that is seeing in PPT?

4

What is the correlation matrix of the cancer volatilome with other traditional classes of features such as from histopathology, imaging, or genomic approaches? Are poorly differentiated tumour cells or cells closer to the brink of EMT more likely to give out VOCs that would be classified by the well-trained canine as a cancer percept?

5

Who is writing a philosophy/framework for vibe-coding? What are some underrated styles and techniques? Where do humans clearly still have the edge — now, and in 6 months?