ceo @zocomputer - prev: founding eng @substack, founding eng @ fin assistant, early team @venmo

brooklyn
i've verified that jev is terrible at chess rob.zo.space/jev-chess
1
2
242
interestingly if you give it the all legal moves decision architecture it is pretty good in the opening because it has memorized so much theory but then after move 10 it is super dumb again
2
69
windows open at the office today, bk is very pleasant vs manhattan
5
26
1,019
e.g. - a leading AI company concludes that releasing a new model has a 10% chance of ending civilization. as the board motions to stop the release, the model produces a persuasive report showing that not releasing it leads to a 14% chance since a less responsible competitor will get there first. set at an emergency summit in luxurious ranch at jackson hole, peter sellars plays the ceo, secretary of war, and lone researcher who solves rsi
who's going to be this generation's sellars+kubrick and make the movie we all need right now?
2
6
353
who's going to be this generation's sellars+kubrick and make the movie we all need right now?
1
3
672
this whole thing with "ai gonna kill us" mostly misrepresenting ai as a big maligned entity - isn't the basic fear everyone thinks others are dumb, careless, misanthropic and we're afraid of giving other people leverage? from rogue actors to coordinated organizations, we're mostly afraid of each other, all leveraged to the information and automation maxx. not so much the paperclip machine or some alien supermorality
11
365
still so many good domains out there
1
8
435
clem one of the good ones that everyone wants to see win
Super happy to share our intention to join forces with NVIDIA in a $12,930,300,000 acquisition 💛💚 10 years after starting Hugging Face, open-source AI is at an inflection point. Thanks to the community, we’ve shown that it can be a complement, and even an alternative, to closed-source APIs. But for it to happen at larger scale, it needs more compute, more support, more collaboration and more visibility. That’s why we went to talk to Jensen, who offered to do exactly that with us. In addition to doubling down on NVIDIA’s massive contributions to open-source AI (I called them the “King of American open-source AI” earlier this year), they’ve committed to strongly supporting Hugging Face and our mission while keeping the platform open, independent and compute agnostic. The founders and the team are all staying to keep pushing this mission forward. Together, we think we can make open source the default way to build AI, with the goal of empowering 100 million AI builders to own their intelligence rather than rent it. Excited about the next 10 years! 🤗🤗🤗
1
13
498
in 2027 anthropic will transcend their claude product line with a new kafka line and `anthropic/kafka-castle-8` will be a big hit with developers
1
3
217
obviously with "recurrent depth" latent space reasoning tokens
1
102
The Dead Leshy Theory of the Mature Internet
2
82
some day soon you/your agent should be able to have any number of omarchy (or any other distro) instances in your zo-net the best experience today is not bad - 5k streaming over UDP at 20ms latency to a node in us-east with an nvidia L4, can be serverless, ~30s boot and ~$0.70/hr while you're using it. few bucks/mo for storage @zocomputer @dhh
5
2
24
3,644
There is this guy in our neighborhood that rides around an electric scooter blasting italian opera every afternoon, which reminded me of this film / doc combo, which i assigned as homework for the team
6
300
still every successful engineering interview i'm seeing these days
Cracking the Coding Interview (2026)
1
12
993
under-appreciated tweet
6
1,064
everyone is now thinking about agents (harnesses, context, loops, etc) but i feel like very little builds on intuitions from actual agency theory. we have all this agent-side focus but the original problem is - how do you align an agent with a principal under information asymmetry. all of that led to principal side objects like contracts, monitoring, and incentive design. so all this (nominally) agent side tooling and the principal side tooling is always... a text box
7
2
69
4,535
i guess why it matters is that we probably would have gotten to self-improvement loops earlier if we factored the concepts more "correctly" - it would have made us develop the ideas around intent, evaluation, and revision more "natively" in these systems
1
10
395