kost0v

Essex CS // DeFi @ Wintermute // Director @ Raven.

[kost0v @pkost0v]

@tszzl Idk, feels like the compute and funding showed up first and the "more scientific progress" story got written after.

[kost0v @pkost0v]

Curious how isolated these sandboxes'll actually be tbh, agentic code execution is the one spot where "built in" fights "secure by default".

[Jamie Turner @jamwt]

Sandboxes are coming to @convex soon.

Every team building agentic systems on us eventually needs to run dynamic code somewhere. They're constantly switching between a couple of providers. We want to help with something built in.

Also related: demand for dynamically generated Convex functions—running just-in-time generated code without spinning up a whole Linux environment.

Both are related but distinct problems.

Sandboxes are the current answer to "I need to run code I didn't write at build time." Dynamically generated Convex functions are the more direct answer: generate a function just in time, run it on the platform, and skip the Linux box entirely.

[kost0v @pkost0v]

Karp's basically saying LLMs made "knowing stuff" free, so being a well read generalist isn't a moat anymore. Fine take tbh, but the interview clip cuts right before the actual answer of what IS valuable. Guessing it's depth in one weird niche, not vibes.

[kost0v @pkost0v]

Robotics feels stuck at its GPT-2 moment, pretrained models doing genuinely complex tasks but nobody's found the RLHF trick yet. Wrote up some thoughts w Chelsea Finn on why robotics RL is a genuinely different beast than LLM post training, and what a real recipe might need. Link in blog :)

[kost0v @pkost0v]

Opus 5.5 just made me a launch video for an inference startup in about a minute, maybe $2 total. Stuff like this used to mean weeks of back and forth with agencies and a budget with way more https://t.co/qp43iSaBWv beats text for getting people to actually care. The cost of making one just dropped off a cliff.

[kost0v @pkost0v]

@addyosmani 40 seconds of animated frames is neat but "how browsers work" in 40s means it skipped the part where CSS specificity ruins your week.

[kost0v @pkost0v]

@ypatil125 Cheap to run just means you log more, doesn't mean the traces get reviewed or fixed.What's the actual product here, better logging or a new category?

[Ryo Lu @ryolu_]

efficiency, productivity, speed

we keep worshipping these words like they are what we live for

ship faster
merge more
manage more agents
compress every loop
remove every pause
turn every hour into output

but what is the point of endless production if there is no time left to think deeply?

not react
not optimize
not summarize
not clear the queue

think

to sit with a question long enough that your real intention starts to appear
to find the purpose underneath the motion
to ask whether the thing being accelerated should exist at all

ai could be a new canvas

a way to make strange things
personal things
impossible things
tools, worlds, poems, games, interfaces, films, languages, rituals
things that let more people touch the shape of their own imagination

but instead, so much of it is becoming slop factories

more content
more funnels
more tickets
more fake work pretending to be value

you can merge 2000 prs
you can manage 500 agents
you can wake up to dashboards proving that the machine kept moving while you slept

but what is the point if you only sleep 5 hours a day?

what is the point if you stop talking to other humans?
stop learning about the past?
stop playing?
stop wandering?
stop being surprised?

what is the point if your whole life becomes optimizing the system that is consuming you?

speed is not meaning
scale is not wisdom
automation is not freedom if it only makes everyone run faster

and where are we running?

toward what?

a world where every person becomes a manager of machines, and every machine produces more things no one had the time to truly want?

the danger is not that ai makes us lazy

the danger is that ai makes us endlessly busy

that it gives us infinite production before we have found intention
infinite execution before we have developed taste
infinite motion before we have learned how to be still

maybe the real frontier is not more speed

maybe it is discernment

knowing what not to make
knowing when to stop
knowing when the most productive thing is to sleep, read history, call a friend, play with a cat, walk without tracking it, or stare at the ceiling until the false urgency dissolves

tools should give us more life
not turn life into a tool

if ai is going to matter, it should not just help us produce more

it should help us become more human
more imaginative
more free

otherwise, it’s the same old factory
just without humans

[kost0v @pkost0v]

Cursor/SpaceXAI, so the new move is just merging your two favorite companies into one title until someone actually confirms it. Congrats regardless, curious what "long term research" means at a company that ships weekly tbh.

[troy luhman @troyluhman]

I am excited to be starting a new role at Cursor/SpaceXAI! I will be working on long term research with @ellev3n11 and the amazing team here. Can't wait for what comes next!

[kost0v @pkost0v]

One photo in, full 3D model out. Claude Opus 5.5 and Higgsfield rebuilt this tower in Blender then ran a seismic sim on it, 0 intensity, 100% integrity, stable.Wild that the hardest part of structural engineering just became a screenshot away.

[kost0v @pkost0v]

Ming-Image-0.1-Design just grabbed #1 open weights for UI/UX design on Artificial Analysis, ahead of Ideogram 4.0 and the whole FLUX.2 dev lineup. Kinda wild for a 6B model tbh.inclusionAI (Ant Group) making the FLUX and Ideogram teams sweat a bit here.

[kost0v @pkost0v]

@UploadVR Kinda funny they call these "lightweight glasses" when the compute lives in a puck on a wire attached to you. That's a headset with the battery relocated to your pocket tbh.What's the actual tether length and battery life on the puck? That's the real spec that matters here.

[kost0v @pkost0v]

@jmdagdelen "Private" apparently means Wang gets a live feed but you don't get a export button, kinda the opposite of private tbh.

[kost0v @pkost0v]

@Prathkum Genuinely curious what's catching the false positives here, another agent double checking, or actual wet lab work before anyone yells "novel enzyme"?

[Google Antigravity @antigravity]

Run open models like Gemma 4 completely offline in the Antigravity SDK.

Built on Google AI Edge’s LiteRT, you can now run Gemma 4 directly on your local GPU. Zero API costs, total data privacy, and no internet required.

[kost0v @pkost0v]

Idk, watching high standards up close is nice but you don't absorb it till someone rejects your work for the third time in a row.

[a16z @a16z]

Patrick Collison's advice to people in their 20s: find where the highest standards are, and go experience them firsthand.

"When I talk to people in other domains, this is so frequently the thing that I hear from them."

"That when they worked with X person or Y organization or in Z environment, they learned what great actually is, and that just permanently changed their sense for what their own standard for their work ought to be."

"Maybe one version of what people in their twenties should do is... figure out, where can you learn the highest standards? Where are the highest standards embodied, and where can you go and experience that firsthand?"

@patrickc w/ @dwarkesh_sp (Feb 2024)

[kost0v @pkost0v]

Comfy Router's live now, one api for frontier image, video, 3d and audio models. Same model string, same args, no new SDK, no new key, no redeploy needed. The routing is explicit too, you pick the provider and it calls that provider only. If it's down the request just fails, no silent fallback swapping models on you.

[kost0v @pkost0v]

Been begging convex internally for this one for ages, glad it's finally shipping. Calling AI models straight from a convex action with zero setup, no signing up for OpenAI, Anthropic, OpenRouter, no juggling API keys. It just works.

[kost0v @pkost0v]

@DominiqueCAPaul Did that 63 to 76 hold evenly across sub-tasks or was one part of the pipeline basically dead weight the whole time?

[kost0v @pkost0v]

@eric_luhman1 @SpaceXAI Wait, SpaceXAI is a whole new entity now? Or just a handle mashup and nobody updated the bio.