

Presenting Sponsor
+ Become the presenting sponsorPresenting Sponsor
+ Become the presenting sponsorPresenting Sponsor
+ Become the presenting sponsorSilver Sponsor
+ Become a sponsorSilver Sponsor
+ Become a sponsorSilver Sponsor
+ Become a sponsorSilver Sponsor
+ Become a sponsorSilver Sponsor
+ Become a sponsorNomination created on September 22, 2026
Get Diogo Almeida on All-In with Chamath, Jason, Sacks & Friedberg!
Champion Hub
Every way to move this nomination, in one place.

Sane + 🌶️ takes in an insane AI world... AI capabilities researcher: co-created RLHF/ChatGPT @ @openai now trying to right the wrong 🤭 (ceo @typesafeai)
Industry veterans, degenerate gamblers & besties Chamath Palihapitiya, Jason Calacanis, David Sacks & David Friedberg cover all things economic, tech, political, social & poker.
Diogo Almeida is the founder of TypeSafe AI and previously worked at OpenAI on InstructGPT and the research that helped turn language models into useful assistants.
Now he’s questioning one of the basic assumptions of the AI boom: if these models are already so intelligent, why have they automated so little actual work?
TypeSafe’s answer is Jev — a different kind of AI model built around decisions rather than conversation. Instead of generating paragraphs of text that software then has to interpret, Jev takes in context and returns typed decisions with probabilities that software can act on directly.
The idea seems to have struck a nerve. Within 24 hours of appearing on Vercel’s AI Gateway, Jev was being used by nearly 13% of paid teams — more than twice the adoption of any previous model launch on the platform.
That makes the conversation bigger than “here’s another new AI model.”
Diogo is effectively making the case that the next major wave of AI may not come from making chatbots endlessly smarter. It may come from making intelligence cheap, fast and reliable enough to disappear inside ordinary software.
Sep 22, 2026
Diogo Almeida — co-inventor of InstructGPT and RLHF, now CEO of TypeSafe AI — has been nominated to appear on All-In with Chamath, Jason, Sacks & Friedberg. Add your support to make it happen.
Sep 22, 2026
On September 15, 2026, Almeida launched Jev and TypeSafe AI came out of stealth, closing a $40M seed round — exactly the kind of contrarian AI infrastructure bet the All-In besties dig into.
Champion Hub
1 person wants this. Add your vote.
0 of 7 done
Record a short video telling everyone why this dialogue should happen. Your clip gets featured right here.
What's a Key Catalyst?
A Key Catalyst is someone whose voice carries trust. By sharing or supporting this nomination, they help ensure the right conversations actually happen.
Rally support from known voices who can help accelerate this nomination

Venture Capitalist and Podcast Host
Why Their Voice Would Be Valuable
Chamath focuses heavily on macro trends, transformative technology, and enterprise-level shifts in computing.

Tech Investor and Podcast Host
Why Their Voice Would Be Valuable
As a prominent tech investor and co-host, Sacks closely tracks enterprise software and how new developer architectures change the software landscape.

Investor and Podcast Host
Why Their Voice Would Be Valuable
Friedberg brings a science and data heavy lens to the All-In table and interrogates the technical viability of frontier technologies.
Next goal: 5
Next goal: 5
Be the founding pledge.
One pledge opens the board for Diogo Almeida.
Charged only if the episode is published within 90 days of your pledge.
Pledges are not tax-deductible. dLogos is not a charity, and it directs your earmarked contribution as your agent.
Jev crashed on launch day from demand — what's the actual bottleneck keeping you from scaling, and is it compute, cash, or something about the RLCD training process that doesn't parallelize the way standard LLM training does?
Sacks and Friedberg will immediately want to know if the launch chaos was a good problem or a sign of architectural fragility at scale. Diogo has shared the excitement but not the technical constraint — this forces a real engineering answer from a founder who knows where the bodies are buried.
Synthetic Data Only: The Bet That Changes Everything
Diogo trained Jev entirely on synthetic data and calls it 'one of the best bets I've ever made.' The All-In crew has debated data moats ad nauseam — Diogo is a live case study of whether you can build a frontier model without scraping the internet, and that changes every assumption Chamath and Sacks have made about data as a competitive advantage.
Suggested
RLHF's Original Sin: Did Human Feedback Break Autonomous AI?
Diogo co-invented RLHF, the technique that made ChatGPT possible, and now publicly argues it's the reason AI can't operate without a human in the loop. No one on the pod has ever interrogated the foundational training choice that produced the current AI paradigm from someone who actually made it.
Jev crashed on launch day from demand — what's the actual bottleneck keeping you from scaling, and is it compute, cash, or something about the RLCD training process that doesn't parallelize the way standard LLM training does?
Sacks and Friedberg will immediately want to know if the launch chaos was a good problem or a sign of architectural fragility at scale. Diogo has shared the excitement but not the technical constraint — this forces a real engineering answer from a founder who knows where the bodies are buried.
Synthetic Data Only: The Bet That Changes Everything
Diogo trained Jev entirely on synthetic data and calls it 'one of the best bets I've ever made.' The All-In crew has debated data moats ad nauseam — Diogo is a live case study of whether you can build a frontier model without scraping the internet, and that changes every assumption Chamath and Sacks have made about data as a competitive advantage.
Here's a 45-second TL;DR on Jev. I find the core idea beautifully simple, but the video made it really hard to understand. Hope you find it helpful.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
JEV is INSANE. We gave it 700 high-intent leads and personalised outreach messages. In 40 seconds, it predicted how each message would perform, assigned a confidence score and detected lead-message mismatches. All for just $0.09. JEV can also score leads, analyse buying
Jev is incredible. I'm scared people are using it wrong.
Jev has completely replaced ChatGPT, Codex, Cursor, Claude Code, iMessage, Spotify, Ford, CVS, and McDonalds for me. It really is the everything model.

Michael Saylor on The Joe Rogan Experience335 votes
Anand Mahindra on WTF is with Nikhil Kamath98 votes
Vaclav Smil on Dwarkesh Podcast27 votes
Leemon Baird on Lex Fridman Podcast33 votes
Elon Musk on Relentless178 votes
Richard Heart on The Joe Rogan Experience1,195 votes
DataRepublican (small r) on PBD Podcast65 votes
Karl Friston on Dwarkesh Podcast22 votes
Alexandria Ocasio-Cortez on This Past Weekend w/ Theo Von7 votes
Robert F. Kennedy Jr on Huberman Lab3 votes
Lily Tang Williams on The Joe Rogan Experience894 votes
Alexandria Ocasio-Cortez on The Tucker Carlson Show7 votes
Nicki Minaj on Real Talk with Zuby6 votes
ZUBY: on Chrissie Mayr Podcast1 vote
Greta Thunberg on Impaulsive with Logan Paul2 votes
Timothée Chalamet on Armchair Expert with Dax Shepard2 votes
Why Their Voice Would Be Valuable
As one of the core All-In hosts and an active angel investor, he frequently dissects startup growth and early stage AI breakthroughs.