@stratechery@benthompson Point at end about sharing in economic benefits similar to Permanent Fund Dividend in Alaska (where I’m from). The logic is: the “people” own public land, not the government, and therefore should directly receive the benefits of its extraction (oil)
@_AndrewCallahan@_AndrewCallahan do you think Roush makes it that far? Feel like I hear on social media/podcasts etc he might rise in 2nd round. Curious what you’re hearing/seeing
@jameszhou02 I wrote a piece about what software cos are defensible in AI era and which aren't. Formula is: "how easy is it to verify correctness" x "consequences of getting wrong":
https://t.co/tK2d8ojcWY
This is a shining example of exactly what I mean
Coding agents make it really tempting to work on everything, but the code itself is only a representation of how thoughtful the inputs are.
The hardest part has never been coding, but rather deciding what's actually worth doing
https://t.co/BMnFr29f8m
@andrew__reed An underrated structural reason imo: at big tech, (output verification difficulty × risk if wrong) is exponentially higher.
Big company devs scared to merge agent changes, while startups with less to lose can be AI first. Move fast & break things more true than ever
@stratechery@benthompson Re the harness being differentiator, how do you reconcile that with harnesses being software, which is easier than ever to copy. In my experience, as soon as new best practices shared, coding agents can implement it practically overnight
.@joinkaizen just crossed a major milestone: 1,000,000 workflows automated.
That represents over 4 million browser minutes — roughly 70,000+ hours, or more than 8 years of operational work automated on Kaizen.
Btw, @joinkaizen created this image for me!
Let me explain what just happened, because I don’t think people realize how INSANE this is.
> Cortical Labs put 200,000 real human brain cells onto a silicon chip and trained them to play Doom in just one week.
> Each CL1 system costs $35,000.
> A rack of 30 units consumes only 850–1,000 watts combined.
> The human brain operates on 20 watts.
> Large AI training clusters burn through megawatts.
>Backed by In-Q-Tel.
115 units began shipping in 2025.
> Cortical Labs is selling “Wetware as a Service” through Cortical Cloud, letting developers deploy code remotely to living human neurons with no lab required,
> priced like a software subscription but powered by real brain cells grown from adult skin and blood samples.
> it isn’t about gaming, it’s about biological computing that could eventually outperform traditional silicon in energy efficiency and adaptability.
This is getting really scary and we’re still at the very beginning.
In the “is software dead?” debate, the idea that customers will just build everything themselves misses an important nuance in the build vs. buy calculus — and that nuance is what determines which companies are actually defensible.
My take:
https://t.co/nz0nDPqxoq
2 weeks ago, we rebuilt our entire product.
"Browser automation" fell short of our mission to eliminate all repetitive knowledge work.
The new Kaizen is the ultimate digital employee: always on, extremely capable, continually learning.
Sign up for access in the tweet below.
@amasad@GavinSBaker@amasad curious how you reconcile this with @Replit. Either there’s some level of product differentiation companies can build into agents, or everything just converges into Claude Code. I.e. every agent company fighting a losing battle against model providers
Humans invented software to free ourselves from manual work.
Somehow, we still waste billions of hours clicking across websites.
Today, @joinkaizen launches to fulfill that original promise.
Announcing our $4M raise from @8VC and the public launch of the Kaizen platform.