Pharmacist. Pharma dev and manufacturing. Writing on mRNA, longevity, tech transfer and AI agents, with sources. Founder @PureMonograph @regentforge @ruvanotech
Vancouver
Joined June 2009
- Tweets3K
- Following2K
- Followers555
- Likes5.9K
Pinned Tweet
Most AI tools in this industry answer every question.
That’s the problem.
If a citation is real but the claim is wrong, the tool should say so. We had a generated pharma report attach the wrong trial to a real identifier. The title of the right trial was three lines below.
Checking that a paper exists is not checking that it supports the sentence.
That’s the work.
When a post reports news and the source later denies it, label the original post. "Denied by source." Right on the post. The false version keeps the reach, the correction gets scraps. Flip that. @X
2026 Midterm forecast, Election Day pre-results. House: D favored (~95% in markets), Sabato 239–196. Senate: D favored (~75–80%), Sabato 53–47. Key flip path: NC, ME, TX, OH, AK, KS. Forecasts, not results. #Midterms2026
How rare and beautiful it is when someone can see around the corner with enough first-principles clarity to actually build the thing. Not just sketch the idea—execute it at scale.
Most people stop at “that’s impossible” or “that’s too hard.” A few ask better questions: What are the real constraints on power, cooling, and mass-to-orbit? What happens when launch costs fall by another order of magnitude? What becomes obvious once you stop treating the atmosphere and the day-night cycle as fixed facts?
Then there are those whose only contribution is to slow it down. Not because they’ve found a better path, but because progress itself feels threatening. History is full of them. They rarely win, but they do waste time.
The people who can both see it and ship it are the ones worth paying attention to.
Gavin Baker (@GavinSBaker) on orbital compute:
"The most important thing in the next 3-4 years is data centers in space.
In space, you can keep a satellite in the sun 24 hours a day. So you don't need a battery. Space cooling is free.
In every way, data centers in space, from a first principles perspective, are superior to data centers on earth."
"@elonmusk operates two of the largest coherent GPU clusters in the world, SpaceX is responsible for over 90% of mass to orbit and SpaceX operates the largest satellite constellation in the solar system.
More than 10 years later, no other company or country can consistently land and reuse orbital rockets."
"Starship should be able to lift 10 to 15 megawatts to sun-synchronous orbit per flight."
"Starship is going to change the space economy in ways we cannot imagine.
Particularly if regulation becomes a constraint to data centers, none of it's going to matter. You're going to sell as much orbital compute as you can make."
A collection of Gavin's quotes from his conversations on Invest Like the Best (EP.451 and EP.473) and his own posts on X.
Kayhan Moayeri retweeted
This is why I warned you to prepare last week.
China is quietly buying up its own stock market again. Their biggest tech companies just suffered 12 straight weeks of losses and Beijing is stepping in for their third massive bailout this year.
Now look across the global map. France can't pass a budget. Poland's government is locked in a fierce power struggle, on the brink of turning into a banana republic. China is propping up its failing AI stocks. And right here in America, our entire stock market is resting on a tiny handful of companies that built their entire future on AI. And they're not turning a profit yet.
A year ago, I recorded a special for Torch Insiders about "capital controls." Most people have no idea what that term means, but if you want to understand what’s happening right now, you MUST know what it is.
I gave you the blueprint a year ago, and every word of it is coming true today. It’s called Capital Controls: How to Survive Financial Collapse like the Wealthy Elite.
Tonight at 7 PM ET, I’m releasing that entire special to the public for free.
GDPx's best model scores 12%. In finance, every model scores 0 under binary grading.
That's the case for agent drafts plus human approval, not autonomy. An agent that passes half the checks is a fast associate, not a signer.
newsletter.semianalysis.com/…
Voceryq, Oct 30: "the first for a plasmid DNA-based immunotherapy in the U.S."
Accelerated. FDA: "Seven (27%) out of 26 patients who had at least three baseline surgeries did not require any surgery intervention during the first year after treatment."
fda.gov/news-events/press-an…
FDA's draft guidance on individualized gene editing is really a CMC document.
One BLA, many patient-specific variants, if each stays inside a pre-set "Individualized Feature Space" and passes lab safety and potency tests.
The envelope is the product.
hhs.gov/press-room/fda-cms-a…
Kayhan Moayeri retweeted
SAMPLE (fictional) approval card, small cleaning company:
Came in: Website chat, Thursday 8:30 PM. Move-out clean, 2-bedroom, wants it done before the 31st.
Drafted reply: "Thanks for reaching out. We have openings on the 29th and 30th. Which day suits you, and is the unit empty by then?"
Flagged for you: They asked for a price. No price was given.
Nothing goes out until you say yes.
ruvano.ca
#Cleaning #SmallBusiness
Kayhan Moayeri retweeted
Holiday hours are coming. Write them down once, then put the same note everywhere customers look:
- Your website
- Your Google listing
- Your email auto-reply
- The voicemail greeting
One version, copied four times, means nobody shows up to a locked door.
ruvano.ca
#SmallBusiness
Kayhan Moayeri retweeted
Question for owners: a quote goes quiet. How many times do you follow up?
A) Never
B) Once
C) Two or three times
D) Until they say no
There's no wrong answer, but knowing your number makes the follow-up easier to hand to someone else.
ruvano.ca
#SmallBusiness
Enveda ran Claude over 1,061 pages of batch documentation. It caught all 11 issues human reviewers found, plus 7 they missed. Review time: 25+ hours to under 3.
AI's first real payoff in pharma is CMC, not molecule design.
anthropic.com/research/early…
Longevity's next breakthrough might be boring: better safety data.
You can't prove something is safe for healthy people to take for 20 years when most adverse events still live in free-text notes nobody systematically reads.
Small businesses don't need an AI assistant that sounds smart. They need one that shows its work.
Every answer it gave, every booking it made, every message it handed to a human, logged where the owner can check it in 30 seconds.
That's the bar.
Agentic AI flagged immune-related adverse events from cancer immunotherapy in clinic notes. Nature Cancer, today.
Retrospective F1: 0.92. Three months live: 0.72-0.79.
The drop is the real headline. Ask every health AI vendor for the live number.
nature.com/articles/s43018-0…
SAMPLE (fictional) log, small property management office:
8:12 AM | Website chat | Tenant reports leaking tap, Unit 4 | Reply approved, passed to you to book the plumber
9:40 AM | Email | Owner asks for September statement | Draft ready, waiting for your OK
11:05 AM | Email | Rent reminder, Unit 2 | Drafted, waiting for your OK
2:30 PM | Website chat | Viewing request | Passed to you
Every row shows what came in, what was drafted and who decided.
ruvano.ca
#PropertyManagement #SmallBusiness
Longevity has a measurement problem before it has a drug problem.
Until we agree on endpoints regulators will accept, most "anti-aging" results are biomarker stories, not outcomes.
Most small businesses don't lose leads to competitors. They lose them to voicemail.
Before spending another dollar on ads, fix what happens in the first 60 seconds after someone reaches out.
Aging clocks will matter to regulators when a change in the clock predicts a change in outcomes. Until then they are measurements, not endpoints. Which clock is closest?
The science of aging with one of its top researchers, Vadim Gladyshev @gladyshev_lab, and optimism for slowing biological aging
with transcript, open-access, w/ links to his many recent papers
erictopol.substack.com/p/vad…
