Co-founder & CTO @ United. Saving humanity from boring work.
New York, NY
Joined April 2022
- Tweets513
- Following1.6K
- Followers1.8K
- Likes3.3K
Another Trojan horse: they'll keep building specialized AI models and agents for each vertical and selling them directly to enterprises, all while making those same SaaS companies offers too good to refuse. Once the models are safe enough for the mass market, they'll replace the whole stack.
Any $10B+ SaaS company that doesn't own its entire stack, models included, will be a dead man walking within 1–2 years.
What Jev's question type do you use to identify claim boundaries in text? I’m exploring Jev for similar use-case, but the docs suggest it’s either unsupported or inefficient form of brut-force: I assume you’d need to try different text spans and ask Jev whether each captures the full claim, only part of it, or extra words.
AI systems are getting more powerful, and they're increasingly being used to build the next version of themselves. We want to illuminate that progress for the public.
Today, we're sharing three measurements that help track AI development:
1. How much AI R&D is done by AI.
2. How well AI agents are overseen.
3. How compute is allocated.
We provide a snapshot of these metrics from inside Anthropic. Any frontier developer could publish the same measures, and third parties could verify them.
As the world considers pacing the frontier, we should do everything possible to minimize the gap between what frontier labs know and what the public knows. This means better measuring the development of AI, publishing our findings, and giving society an opportunity to decide how to use this information.
Read the full post and methodology: anthropic.com/institute/meas…
One ninja hack with Astra + browser control.
Instead of watching it do QA (slow and inefficient), ask it to record a video with timestamps for each test scenario.
Makes reviewing UX changes A LOT easier.
