Rex Salisbury

1/ Excited to announce the launch of @Cambrian_HQ 's inaugural $20 million fund to back the next generation of fintech founders! Incredibly grateful for my time at @a16z, and, now, exited to share how I ended up as a solo capitalist by way of community building. 🧵 👇

Are marketing agencies obsolete, or are AI agents just the new black box? Seema at @a16z breaks down who actually wins when software eats the marketing department. () Are Marketing Agencies Becoming Obsolete? () Will AI Agents Replace the Marketing Team? () Is GEO Going to Kill SEO? () Can We Trust Ads in the LLM Era? () Who Wins the AI Marketing Race? () The End of Expensive Market Research? () How Must the CMO Role Evolve? () What Does True AI-Native Marketing Look Like? () Overrated or Underrated: Direct Mail & SEO? () The Great SF Pizza Debate () Why Are CEOs Still Speaking at Conferences?

honest question -- if you are an LP in a crypto manager, do you want them to stick with crypto or expand scope?

crypto vcs dont want to invest in crypto anymore...

and a reminder that subsidizing demand will not make housing cheaper. - freezing property taxes - home buyer credits - mortgage interest deduction - rent credits etc etc all subsidize demand give ppl more money to bid up price of a scarce asset...price goes up. the answer is supply.

just want to remind you one more time that significantly lowering the cost of housing would solve, overnight, almost every single problem in the country

property tax freezes will make property MORE expensive. a property tax freeze is subsidizing demand. that will get capitalized into prices espc in supply constrained markets. generational theft. we need housing supply. not demand subsidies. demand subsidies are expensive. unlocking supply is free.

@DKThomp proposal -- any senior can freeze their property taxes if they first attend 10 town hall meetings to support more housing. if they want affordable housing for themselves, need to show they want it for other ppl too.

this is getting ridiculous

if raising an A, why would you not do this this? dalton is great, process is fast 🤷‍♂️ as a seed investor I tell all my companies to do it.

@emmymrtin competition is NIMBY boomers who have spent 50+ years blocking new housing. the framing of "new jobs is the cause of housing problems" is not the right framing. suggests the fix is "job destruction". the fix is housing supply.

As OpenAI and Anthropic prepare to go public, San Francisco tech workers making six figures say they cannot compete with the new A.I. elite. Some doubt they can afford to stay.

@daltonc inverse is even more true b/c the above is true!

The world can be a particularly frustrating place for people with no intuition for statistics.

@Mark_J_Perry american hero.

Boom

This is the chart I would show to argue against AI bubble concerns. Yes AI is the fastest growing spend category ever. It's also still, even for the top quartile of firms, fairly small...<1% of total spend per Ramp data. In @azeem's great report on the state of the AI economy.

@cozymaximalist actually it does. bill diversified, but MSFT is still huge. CZ is concentrated in one large company, but that company is tiny compared to MSFT.

The world you were born in no longer exists

@upgradeoptimism claude pulled all of them, they reference where in the app. there are almost certainly errors. but i agree something like this is so much better than 30 seconds of my dr explaining it verbally!

@rexsalisbury Where did you get the accuracy/precision numbers? I’m curious how these have changed over time (as in what is the trendline)

should you spend $950 to get the galleri test for 50 cancers? i vibe coded a dashboard that helps you decide. if you're over 58, I think answer is yes. If you're younger depends on your tolerance for likelihood of false positives. If you are 40 years old, with no prior family history, 84% of the time, if you get a positive result, it is likely a false positive and it would cost $998,000 to detect one true instance of cancer. at age 58 that drops to 50% and $182,000. the thing I'm most excited with about the Galleri test -- it's cost will go down and accuracy will improve! right now it costs $950. that means it's pretty expensive to detect true incidences of cancer (roughly $1 million or more if 40 or younger) if the cost falls to $100 and accuracy, well it just got a lot cheaper to find real cancers! this is vibe coded. very probably something is wrong. I built this in 30 minutes, would love someone to build a better one! check it out and play around w/ it yourself. test-or-not.netlify.app

@afshineemrani vibe coded a tool to help ppl decide if it's worth it for them. x.com/rexsalisbury/s…

I'm a cardiologist. NPR reported this morning on something that could save more lives than any drug I've ever prescribed. One blood test. One vial. Screening for 50 different cancers simultaneously. It's called Galleri. And the FDA could approve it later this year. Right now, we routinely screen for exactly five cancers in the United States — breast, colon, cervical, prostate, and lung. Each requires its own separate scan or exam. For the rest — pancreatic cancer, ovarian cancer, liver cancer, esophageal cancer, gastric cancer, and dozens more — we have no routine screening at all. We find them when symptoms appear. By then, most are Stage 3 or 4. By then, for many patients, it's too late. Pancreatic cancer has a 12% five-year survival rate — because we almost always catch it late. Ovarian cancer: 50%. Liver cancer: 21%. These numbers aren't medical failures. They're detection failures. The treatments exist. We just find the disease after the window for those treatments has closed. Galleri changes the math entirely. Here's how it works. Every tumor — no matter where it is in your body — sheds tiny fragments of DNA into your bloodstream as cancer cells die and divide. These fragments carry specific methylation patterns — chemical signatures that are unique to cancer cells and different from the DNA your healthy cells release. Galleri captures these fragments from a standard blood draw and reads their methylation patterns using next-generation sequencing and AI-driven analysis. The AI doesn't just detect whether cancer is present. It predicts where it's coming from — which organ, which tissue type — with over 90% accuracy in studies. One vial of blood tells your doctor: there's a cancer signal, and it's likely originating in your pancreas, or your lung, or your liver. Your physician then orders targeted follow-up imaging to confirm or rule out the finding. Galleri isn't a diagnosis. It's a precision compass that tells your doctor exactly where to look. The data is building fast. GRAIL has now sold over 475,000 Galleri tests commercially under a special FDA designation. The NHS-Galleri trial — the largest randomized controlled trial of any multi-cancer detection test in history — enrolled over 142,000 people aged 50-77 in England. The primary endpoint — an overall reduction in late-stage cancers — was not met. But by the third year of annual screening, they found a 26% reduction in Stage IV cancers in key deadly types including pancreatic, liver, lung, and gastric. The test detected four times more cancers overall when added to standard screening — catching cancers that would otherwise have been found late or not at all. The U.S. Pathfinder 2 study — 25,490 participants — showed similar positive signals and forms the basis of the FDA submission filed in January 2026. Congress has already acted. The Nancy Gardner Sewell Medicare Multi-Cancer Early Detection Screening Coverage Act passed in February. If the FDA approves Galleri, Medicare will begin covering one test per year starting in 2028. The current retail price is $950. Exact Sciences' competing test Cancerguard is $659. These prices will fall dramatically once FDA approval triggers insurance coverage and competition scales. As a cardiologist, let me tell you why this matters far beyond oncology. Cancer is now the number one killer of Americans over 50. Not heart disease. Cancer. And the patients I lose to cancer are often the same patients whose hearts I saved — patients who survived their cardiac event, optimized their metabolic health, and then received a late-stage cancer diagnosis that nobody screened for because no screening tool existed. I've written on this platform about GLP-1 drugs reducing cancer metastasis by up to 50%. About personalized mRNA cancer vaccines cutting recurrence by 49%. About inflammation as the common root of heart disease and cancer. About AI detecting disease years before symptoms. Galleri is the missing piece that connects all of it. Detect the cancer early — with a blood test. Confirm it with AI-enhanced imaging. Treat it with personalized mRNA vaccines, targeted therapy, and GLP-1 drugs that may slow progression. Monitor response with liquid biopsy in real time. That's not five separate breakthroughs. That's one integrated system of cancer prevention and treatment that didn't exist five years ago — and could be standard of care within five more. The shift from reactive to proactive medicine — from "we found it too late" to "we caught it in time" — has been the central theme of everything I've written on this platform. Preventive cardiology. Advanced lipid testing. Inflammation detection. AI imaging. Gene editing. Galleri applies the same principle to cancer. And it could save more lives than all of them combined. One blood test. Fifty cancers. FDA decision expected this year. Prevention is the new cure. And the science just took its biggest step yet. open.substack.com/pub/afshine/p/…

@pitdesi Amazing. I’ve been wanting one since I hear about an earlier version that cost $20k 17 years ago. Now well under $1k.

A lot of haters for this but I’m all for it Put it anywhere that it’s hard to get natural light

@pitdesi I think about this skylight all the time!

this is true but conflates two things. you career is likely better in the "selfish pursuit of profit". but if you make a lot of money, that does not mean conspicuous consumption is better for the world than effective philanthropy. rockefeller was a capitalist. he spent some of his spoils providing the initial funding for norman borlaug to research wheat rust in mexico in 1944 that work eventually lead to the green revolution (a term coined by USAID who helped fund the green revolution globally...borlaug credited with saving around 1 billion lives). the world was better off w/ Rockefeller as a capitalist. but it was also better off b/c of the charitable things he did w/ his spoils.

@cremieuxrecueil No one wants to hear this, but more lives have been bettered or saved in the "selfish" pursuit of profit than in "selfless philanthropy.

If someone doesn't have a good answer to 'Why aren't you spending all your money on charitable giving? Your money could save lives!', don't expect them to be able to be reasonable about any other matters of policy.