phil

Today, I’m excited to formally announce @MirendilAI with my amazing co-founders Harsh Mehta, Shayan Salehian, and Tara Rezaei! We’re fortunate to work with @a16z and @kleinerperkins, who led our seed round of $200M, followed by a major investment from NVIDIA, among others. Mirendil exists to accelerate science and technology, and through them, to help solve humanity's most pressing problems. Self-accelerating AI R&D is the most direct path to delivering on AI's broader promise, which is why we believe the most important application of AI is AI itself. Get this loop right, and it compounds. It fundamentally changes the rate of progress itself across all domains. We believe this capability should be democratized. It should be used to power all scientific efforts trying to innovate at the frontier. There are far more important problems—and broader ones—than any single lab can take on, so more groups should be able to pursue them. This pulls concentration of power away from a few labs: businesses and science labs can own their AI and infrastructure, keep their margins, and control their own destiny instead of ceding it all to a single AI lab. We’re a small team with a singular focus. Our founding team consists of 20 researchers and engineers from frontier institutions including Anthropic, xAI, Google DeepMind, and OpenAI, united by a passion for science and a drive to build the technologies that move it faster. If you want to build the system that builds systems, join us! @HarshMeh1a, @shayan_, @tararezaeikh

@bneyshabur @MirendilAI @a16z @kleinerperkins incredible team and mission. undoubtably great things to come

you gotta embody your agents

@ClementDelangue quintessential french directness

Shot a video this morning to announce a new collaboration with ...

⚽️

@AnikaSomaia brighter lamps have reached a shibboleth status among the new cultural vanguard of sf to a degree that no other consumer product has honestly remarkable

for a company that makes lamps btw

@AndrewYatzkan the scanner does what to you now

I was the 6th person to get FUCT from the Midjourney scanner I had lunch 10 minutes before and my entire torso was just stomach

@_sholtodouglas this product's provenance makes a hell of a lot more sense than people are giving it credit for

If deployed widely - I bet this will save the US healthcare system at least 100x all of MJ’s profit to date. It’s a great example of how much better someone like David is able to allocate capital than ~the rest of the world. The AI boom should empower a generation of people who understand just how fast we can climb the tech tree, and will dream very, very big. Expect incredible things. When Midjourney Dyson Spheres?

@imjaredz @_patrickhult @jimmyliu @NotionHQ @varadh there is a parallel universe involving several PoDs resulting in a street called "the ocky way"

I can’t believe these madmen actually bought a street in SF

ngl having trouble keeping up with the release cadence lol

self-driving in june 2026 feels like coding in dec 2025

SpaceX has exercised the option to acquire @cursor_ai in an all-stock transaction with the goal of building the world’s most useful AI models. For the past few months, SpaceXAI has been jointly training a model with Cursor, which will be released in Cursor and Grok Build soon. We look forward to working closely with the Cursor team to advance our frontier AI capabilities

@houjun_liu @amelia_f_hardy @jshetaye @DuncanEddy lfg???? LFG!!!

</undergrad> "we should make a bingo card of convos with you" — @amelia_f_hardy "poopmoka" — @jshetaye "so proud of you [for getting into a bar fight]" — @DuncanEddy "you've gained weight" — most family members ggs; back to work

parallel agents of chaos

@cremieuxrecueil if your lab doesn't have an endonym for its people you're ngmi

"Anthropic employees call themselves 'ants'." "OpenAI employees are 'ops'."

@KatieMiller @SpaceX FSAE success is probably one of the strongest early predictors of engineering excellence

“Race cars and rockets are not that dissimilar,” @SpaceX’s Bill Riley, once told an interviewer for the Ivy League school’s magazine. SpaceX executives Mark Juncosa and Mike Nicolls also cut their engineering teeth making student race cars at Cornell in the early 2000s. The Cornell connection at SpaceX is striking, but it also reflects the company’s longstanding commitment to hiring people steeped in practical skills, and not just prowess in the classroom.

@kalomaze datum for what qualifies as "big model smell" changes with time obv but certain tasks in posttrainbench.com are very good at splitting models w/wo bms

i am trying to work on the closest thing possible to a true "big model smell" eval which is to say: something that measures something that clever post training can't trivially gap, and is cheap + topically diverse i can't test mythos for obvious reasons, but... hmm...

@HeinrichKuttler does it outright refuse or just not grok the solution

Fable 5 still struggling with the harder classic analysis question that I'm idiosyncratically attached to.

cool eval

@khoomeik neither one will be the sole limiting factor and the world is not (yet) efficient enough to exploit one to the extent that all efforts go to solving the other

probably an important question rn: is the late 2020s gonna be constrained on chip production (fabrication + packaging) or chip deployment (datacenter + power)? i hear a different take on this every day and i just want a clear answer. any good pieces i can read?