Notes from the field.

Most 'AI products' aren't products. They're UI on top of someone else's model.

On why model providers will keep moving up the stack and compress margins — and why the AI startups that survive will own workflow, data, and distribution in a specific vertical.

Mar 2026

LinkedIn↗

This isn't a downturn. It's a discipline cycle.

On biotech's barbell market and why the question isn't 'can you raise?' but 'have you earned the next dollar?' — a finance leader's take on the 3–5 year window ahead.

Mar 2026

LinkedIn↗

The 409A valuation trap: freezing what should stay fluid

On how 'defensible' 409A methodology can lock in assumptions that still belong in scenarios — and why the cleanest model is often not the most accurate one for milestone-driven biotech.

Mar 2026

LinkedIn↗

Biotech doesn't need more dashboards. It needs renaissance developers.

Coding alone isn't the differentiator anymore — context is. The next decade of biotech leverage belongs to hybrid builders who can reason across science, finance, and AI simultaneously, and use AI to collapse the time from idea to implementation.

Dec 2025

LinkedIn↗

The next era of AI won't be trained on the web. It'll be trained on work.

The open web is nearly tapped out. What comes next isn't more data — it's expert workflows. How scientists design protocols, how clinicians decide to halt a trial, how finance teams model cash runway. That judgment is the new training fuel.

Nov 2025

LinkedIn↗

Customer zero: how CFOs should think about agent investments

Notes from a Robin Washington (Salesforce) chat on the agentic enterprise. The finance angle: tie agent investments to productivity you can bank, then recycle that capacity into growth. Go slow to go fast, reallocate don't just add, chase NPV not only payback.

Nov 2025

LinkedIn↗