Christian Vestergaard retweeted
this is it. 1000% it.
people who think something is too late are usually not founders. one could argue that book business was done and amazon appeared to change the game. blockbuster was the king when netflix was the cd delivering peasant til the empire was overthrown.
again and again when the non founders think the game is over, the founders appear and change the game.
it's an infinity war if you choose to believe it, and the game is always on.
Scott Wu on the nihilism creeping into tech: people saying it's "too late" to build an independent company because the labs will just do everything.
His take: those people aren't founders. Founders are rationally optimistic with zero evidence they'll succeed.
If you think coding agents look like this forever, you're wrong. There are 10 more generations of this product to build.
there's so much moat in shipping with extreme intention now that LLMs make things abundant. I'm a big opponent to the focus on shipping more PRs and quicker. they are not bad per-se (!), but the metrics tend to attract people with little intentionality, I think.
57
Christian Vestergaard retweeted
The evals are not ok
New NeurIPS paper: Hidden Measurement Error in LLM Pipelines Distorts Annotation, Evaluation, and Benchmarking.
Evals drive deployment, safety standards, and research. Yet we ignore noise from prompt, judge, agent harness etc, & so underestimate error in benchmarking & annotation, & it gets worse with more data.
I'll get to what this means for CIs & how to fix, but first let's talk about how this makes benchmark hacking easier 🧵
It's mind boggling to imagine how we got any work done before LLMs. I'm taking it for granted.
1
3
141
On-policy distillation should be renamed office hours. Everyone would immediately understand the concept.
Christian Vestergaard retweeted
To make something *truly* great, you always have to put in more time and effort than you want.
The good news is most don’t, so if you do, it’s all the more likely it’ll stand out.
It's time for new training regimes in speech. We believe audio-native RL environments with fine-grained rubrics (naturalness x task completion) is part of the answer. We've been doing research on this and can soon release a preview of our own post-training. Truly exciting times!
Christian Vestergaard retweeted
We believe most interaction with AI will happen through speech. But models still talk like machines.
This is a data problem.
Today we're launching @agenticdataco (YC F26): real-world speech data + audio-native RL environments so speech models sound and act human.
I listen to "Duel of the Fates" while I work, you listen to lofi beats - we are not the same.
1
5
265
Don't sleep on this release! We lack great open-weights full-duplex models. Nvidia's VoiceChat is the only open model that can think, act and talk at the same time - like we see the closed models do. But the model is too small to work in production. Duo could change this.
Today, we're introducing Duo.
Duo is a full-duplex conversational AI. It listens while it speaks and hears your voice directly instead of a transcript, so it reacts to pauses, a quick "mm-hm" and interruptions as they happen.
2
39
2,651
Christian Vestergaard retweeted
Today, we're introducing Duo.
Duo is a full-duplex conversational AI. It listens while it speaks and hears your voice directly instead of a transcript, so it reacts to pauses, a quick "mm-hm" and interruptions as they happen.
looks like agents are emerging as a new direct channel next to email, sms, and mail. will we all have our own personal agent address soon? just like we all have a gmail or hotmail, it will be muses and groks and instincts.
2
159
Is eventmaxxxing (assuming only smart people) considered following Richard Hamming's advice on being a great researcher is being a visible researcher - or is it pure distraction? Genuinely in doubt.
1
7
374
Too frugal to pay for an expensive launch video - so we found a free content studio and spent $250 on a blackboard.
1
11
280
these guys are smart! delayed consequences is such an alien concept to agents
The economic value of agents will soon come from running whole autonomous businesses, not just from finishing one-off tasks. When you run a business, the result of your own decisions arrive weeks later. You have to notice them, trace them to something you did, and act before they compound.
Today, we're excited to announce SMB Bench, where we measure all the frontier models' ability to observe changes in the environment and adapt accordingly.
2
8
437
a heuristic for spotting great founders: do they pitch/defend their product or ask questions?
2
3
123
Don't be so cool you can't appreciate cool stuff. Bring back some of that earnest childhood wonder, some of that pure unadulterated WOW! Retire your cynicism.
“This is the most exciting time to be into computers”
When I interviewed @dhh at our Omarchy Shenzhen meetup 2 weeks ago what stood out was his absolute enthusiasm - everyone in the room could feel it
😎















