purveyor of paradox
San Francisco, CA
Joined January 2013
- Tweets11.5K
- Following1.4K
- Followers3.4K
- Likes24.7K
real ball knowers know that you could sell a car as an appreciating asset, because the computer inside could theoretically run an llm
a probable mental model for the future is to conceptualize any problem as a block of marble
the goal is to direct a stream of tokens to carve a solution from it as quickly, efficiently and correctly as possible
today we call this a harness
the worlds appetite for inference is infinite
all excess tokens are being spent on securing infra from tokens spent on attacking
cat and mouse games already have and will emerge at every cross section of the economy
the valley is amazing because every ~2yrs the people who made a bag disappear and the people who missed the bag post proof
graphql, rust, mcp, storybook, microservices, db per service, data mesh, data lakehouse
i was wrong about all of it. you needed to be a big company but now that everyone is going to be managing swarms its probably all directionally correct
yeah so thats a banger
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
a while back we worked with 1password on a preview of this that integrated into @trydirector
so excited to see this generally available as a much needed extension for the personal assistant space
