From Devoted to drug design: Annie Lamont on AI's healthcare moment

Key points

Key takeaways from Sourcery interview with Oak HC/FT Co-Founder and Managing Partner Annie Lamont (October ️2026):

AI's inflection point in life sciences. After a decade of investing in AI without seeing it produce developable products or therapeutics, Oak HC/FT believes everything has changed in the past two years. Its 2025 investment in Chai Discovery is building models and products for drug design: antibodies today (around a hundred times more productive than a lab without them), peptides next and small molecules after that.

Doubling down on drug discovery. Healthcare is around ️70% of Oak's activity and life sciences is a deliberate strategic focus: the firm plans to add partners and expand its San Francisco office, reasoning that so much AI development — including life sciences — is happening there.

China is an existential threat. Over the last four years ️50% of the outside R&D dollars pharma spends have gone to China, against not the case five years ago, and ~90% of antibiotic ingredients are produced there; China's universities now produce as many or more patents as America's. Lamont argues that undermining NIH-funded university research and an unprepared FDA compromise the drug pipeline, demanding a "complete re-engineering" of the FDA and its processes.

Devoted Health as the benchmark. She calls Devoted Health possibly the best example of AI applied to healthcare in the world: a decade in the making Medicare Advantage plan, licensed across all 50 states, with its own technology platform, virtual primary-care network and ownership of risk and the member. In the past year it has tripled in size, sharply improved its operating ratio and EBITDA, and an AI front end backed by doctors sits behind the member experience.

Legacy institutions endure — but costs fall. Hospitals and providers are not going anywhere, but AI will lift administrative burden off clinicians: 25–30% of US healthcare costs are administrative, and electronic health records organized the data AI now works with.

Clinicians are augmented, not replaced. Radiologists will not disappear — 30% of radiological images are currently misread — but AI behind them makes misses much less likely; patients already use chatbots that may diagnose better than some doctors.

Surgery enters the AI era. Vision models and robotics will let surgeons see and operate more precisely than today, making Da Vinci look almost brute-force in comparison; AI-guided remote surgery could also narrow rural access gaps where outcomes are dramatically worse.

Read more: Sourcery