@wzhang
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
line cook @zerosetinc
chinese latent space
Joined July 2019
- Tweets1.2K
- Following462
- Followers1K
- Likes4.4K
Pinned Tweet
We raised $5.2m at Zeroset to bridge the gap between frontier intelligence and the trajectories of enterprise workflows.
We believe in a future where companies own the models trained on their own traces.
If this interests you, join us at zeroset.com/careers
Grateful to @darian314 @denise_teng25 @alexiskold @thisiszann for supporting our mission!
Today, we’re announcing Zeroset’s $5.2M pre-seed round to deploy world models into enterprises, co-led by @GradientVC and @2048vc, with participation from Leblon Capital and others.
Frontier intelligence is capable of enormously complex tasks. And yet with an abundance of general capability and mandate, automation still requires enormous handwork. The models and systems that exist today lack an understanding of the world they act in.
Robotics faced a similar problem. Building capable robots required more than compute-scaling on policies; they needed representations of their environments, encoding state and how it changed under actions. Over time, those representations shifted from hand-specified states toward learned models of the world.
What is missing in the enterprise is a general system that represents work and learns from it. At Zeroset, we believe the same paradigm that enabled progress in robotics can be applied to represent enterprise “worlds.”
We open with Nebula, a system that captures the state and traces created as people and agents work across systems, using them to model how an enterprise works and changes over time.
These models contain the decisions, dependencies, and unwritten processes bespoke to a company. They are derived from the enterprise, and they should belong to the enterprise.
Companies should be able to use the best models for their work and train them on their own operating knowledge freely. The underlying system does not need to change.
Nebula is currently in closed research preview. If your team is automating complex work or exploring how to learn from the work your organization already does, please contact us.
We are building the system through which that intelligence becomes increasingly specialized to the enterprise, allowing agents to act through it, and models to learn from it.
If you’d like to join us in building this future, we’re hiring researchers and systems engineers.
Read more here: businessinsider.com/zeroset-…
the only winner post-agi are the coffee shops for sf tech bros until the frontier labs acquire them because they wanted to vertically expand further up the ai stack (jensen huang's 6* layer cake)
if you're consistently paying $1 for a code review you're not doing it right, let alone $5-20 💀
our internal code reviewer running 6.1 sol xhigh btw
running your own internal greptile is usually more expensive than paying for greptile.
usually $5-$20/review to get in the vicinity of greptile quality vs $1/review for greptile.
we have one of the largest dataset of buggy PRs in the world. this means we can know for sure if a particular version of our agent is more or less effective than another.
as a result, we can experiment with hundreds of permutations of models, harnesses, techniques, tools to eek out maximum performance per dollar.
we then pass those savings onto the user.
we can only do this because of the dataset. if we didn’t have it, we won’t know for sure if a new version really can catch every unique one in a million bug.
William Zhang retweeted
Frontier intelligence is getting better at acting. What it still lacks is an understanding of the world it acts within.
For enterprises, that world is not just documents and static retrievable context. It is the enterprise as a dynamic system, including the context and traces that change the static world.
Robotics learned that capable policies require representations of their environments. We think the same will be true for enterprise intelligence.
We’re building Zeroset to create these representations: world models of enterprise state and dynamics, learned from the traces generated by people and agents at work.
Our first system, Nebula, captures this operating state so agents can act with deeper context and models can learn from the work performed inside the enterprise.
Today, we’re announcing Zeroset’s $5.2M pre-seed and Nebula, now in closed research preview.
Today, we’re announcing Zeroset’s $5.2M pre-seed round to deploy world models into enterprises, co-led by @GradientVC and @2048vc, with participation from Leblon Capital and others.
Frontier intelligence is capable of enormously complex tasks. And yet with an abundance of general capability and mandate, automation still requires enormous handwork. The models and systems that exist today lack an understanding of the world they act in.
Robotics faced a similar problem. Building capable robots required more than compute-scaling on policies; they needed representations of their environments, encoding state and how it changed under actions. Over time, those representations shifted from hand-specified states toward learned models of the world.
What is missing in the enterprise is a general system that represents work and learns from it. At Zeroset, we believe the same paradigm that enabled progress in robotics can be applied to represent enterprise “worlds.”
We open with Nebula, a system that captures the state and traces created as people and agents work across systems, using them to model how an enterprise works and changes over time.
These models contain the decisions, dependencies, and unwritten processes bespoke to a company. They are derived from the enterprise, and they should belong to the enterprise.
Companies should be able to use the best models for their work and train them on their own operating knowledge freely. The underlying system does not need to change.
Nebula is currently in closed research preview. If your team is automating complex work or exploring how to learn from the work your organization already does, please contact us.
We are building the system through which that intelligence becomes increasingly specialized to the enterprise, allowing agents to act through it, and models to learn from it.
If you’d like to join us in building this future, we’re hiring researchers and systems engineers.
Read more here: businessinsider.com/zeroset-…
William Zhang retweeted
Zeroset is building temporal memory for agents! Excited to see them emerge from stealth with a $5.2M pre-seed led by us @GradientVC and @alexiskold - read more from @BusinessInsider: businessinsider.com/zeroset-…
William Zhang retweeted
Absolutely thrilled that @2048vc backed this super special team and work with @darian314 at @GradientVC
Zeroset is building temporal memory for agents! Excited to see them emerge from stealth with a $5.2M pre-seed led by us @GradientVC and @alexiskold - read more from @BusinessInsider: businessinsider.com/zeroset-…
i feel like even a "thanks, done sir" would be better than meat proxying onto their gracious human-driven pr here
clean code guy @unclebobmartin says he creates good software without reading the code
was curious so I spent an afternoon reading some recent code.
not sure i agree, PR here 🤗
github.com/unclebob/crapper/…
TL;DR: read the code pls Bob
xitter.cf/unclebobmartin/status/…
true that agents bloat tf out of unit tests but its still the best way to set invariants/guard against edge cases into your code
good luck migrating/rewriting existing code without unit tests
tbh just dont get lazy when writing tests. dont vibe them and make sure they actually test behavior and not implementation
IMO the only tests you should have in the age of coding agents are, in order of priority:
1. Full E2E tests. Nothing mocked out at all. Can even be something that runs in prod on test accounts via Playwright or equivalent.
2. Integration tests. Agents make more mistakes as things bleed between data/API boundaries where schemas can drift.
3. Golden tests. This helps ground the code with real examples of data, and can be used as regressions against edge cases.
Unit tests are bloat 99% of the time if you’re using a frontier model since they’re smart enough now to get the implementation (and subsequent iterations) right on the first shot, so they only add bloat at this point.
even with the most beautiful login page,
0% marketshare gained - the “easy to use” cloud hosting platforms are already aws wrappers and any serious company will migrate from there to aws
just stop making cost manager cost money to view
