US company is editing human embryos. Legally. And I'm an investor.

Dear friends -

A company is editing human embryos inside the US. Legally, right now. And I'm an investor.

In June, Origin Genomics received IRB approval to run in vitro gene correction on human embryos at its New York facility. Origin says it is the first company ever granted it. Research only, implantation remains a bottleneck in the US, and that line still matters. But an ethics board in the United States has now looked at germline editing inside a commercial lab and said yes. The science is maturing, and regulation is catching up. That's a potent combo.

I've been arguing for a while that the conception and embryo genomics stack is the least obvious big thing in frontier bio. This is the month it stopped being theoretical.

Things I’m looking at right now:

Another reason I am bullish on this space is that, outside of editing, there are companies on the market already generating revenues today - offering services such as polygenic embryo selection. This was not the case 1 or 2 years ago. After much research and interviews, I've recently invested in the scientific leader in this space - Herasight. I've linked my X article on their technology, above. I have also previously funded an artificial womb company, still in stealth.
I am actively looking at other companies engineering new products unlocking complementary capabilities, across this stack. Place all these in a context where 130 out of 194 countries are below population replacement levels, 26% of childless couples experience impaired fertility, whilst IVF has crossed 2% of all US births and is not stopping there. I will be publishing my detailed thesis on this space soon - stay tuned!

Outside of embryo genomics, it's hard not to be also bullish on the AI-enabled biotech thesis. The situation here is the mirror image of the former. The capital has arrived with a bang (Isomorphic Labs having raised $2.1B alone!). Yet, 95% of AI drug discovery startups today raise on vibes alone, with no hard data to prove their approach moves the needle where it matters.

Therefore, one has to dig deep beneath the headlines, and be discerning. The question to ask is: 'What data did you generate that did not exist before you started?', 'How hard is it to replicate?, and 'How well does it tie to outcomes, outside of simulation?' Each question gets progressively harder to answer. I’m actively monitoring the space and talking to anyone who has a credible horse in the race.

New investments:

I’ve mentioned Herasight. I’ve also participated in the $8M Series A extension of Kadence Bio - a sexual enhancement company, starting with the development of the first FDA-approved treatment for premature ejaculation. They’re currently wrapping up their Phase 2 clinical trials. 
The company is backed by the billionaire Christian Angermayer -  whom I had on the podcast recently to explain his thesis on human enhancement going mainstream, and what it means for consumer pharma. It can be summarized in one sentence as ‘vanity drugs make more money than AI’ - and it’s a thesis I share. People spend more than $370b a year, cash, on looking and feeling better. No insurer, no formulary, no prior authorization. This is 5x more than all AI frontier lab revenues today. And GLP-1s are just getting started. Btw, Christian’s other company, Atai Beckley, just got acquired by Eli Lilly for up to $3.8b.

Trends aside, I’ve passed up on most peptide companies where branding is the only differentiating factor and no IP or supply chain moats exist. I’ve also passed up on several lab automation startups at risk of falling into the fragile middleware category. Many biotechs claiming to be building a platform, without focusing on proving a single asset. And any startups promising incremental improvement of existing healthcare delivery, instead of unlocking new capabilities.

Portfolio updates:

• Origin Genomics received the first-ever IRB approval to conduct human embryo gene editing experiments, within the US.

• Dognosis reported ~90% sensitivity for early-stage cancer across 20+ cancer types, and opened Phase 3 across 10 hospitals and 10,000 patients.

• Herasight launched PGT-H, cutting intellectual disability risk from 8% to ~4% for first-cousin couples selecting among five embryos.

• Lume Health launched its cortisol pilot study with Equinox in San Francisco, for the first continuous wearable hormone monitor 

• Kadence Bio told BIO 2026 it is ahead of schedule on KH-001's Phase 2a, with clinical proof-of-concept data due in December.

On Accelerate Bio pod

Billions are pouring into AI drug discovery. Zero approved drugs. But that's about to change.

I sat down with the Sam Altman-backed Retro Biosciences Head of Applied AI Rico Meinl, to talk protein engineering, virtual cells, autonomous laboratories, AI co-scientists, why generating “black-box data” may matter more than building larger models, and what's required to turn AI breakthroughs into drugs that actually work.

The ask

• If you are building in embryo genomics, direct-to-consumer bio, AI pharma, or human capability enhancement - I want the narrow provable bet, not the grand vision slide. Reply to this email or DM me on X.

• If you're an investor who wants the conception stack thesis when it's out, reply to this email and I'll send it.

• If you disagree with any part of my thesis and can make a rational argument to support it, I’d love to hear from you too. Will reply to all.