Build: Automating the Built World
From a seed conversation about shipping hundreds of thousands of buildings to compressing weeks of technical due diligence into hours—why we backed Build to construct the digital substrate for the built world.
From a seed conversation about shipping hundreds of thousands of buildings to compressing weeks of technical due diligence into hours—why we backed Build to construct the digital substrate for the built world.
The market for computers' silicon brains has quietly grown from the size of Major League Baseball to the size of the global cement industry, and it is doubling. Computing is bigger than you think, and getting bigger faster than you can update your intuitions about how big it is.
From a leaky seagrass-planting prototype in Dublin to the autonomous platform for the entire ocean—how Ulysses is building the Ocean Company.
"Artificial intelligence" is a technical area, but "intelligence" is not a technical term. When AI people say "intelligence", "AGI", or "ASI", they are vibing just as much as everybody else is. This leads to much misunderstanding and confusion. The technical area that can shed light today on AI's capabilities is computational complexity.
The history of all hitherto existing companies is the history of management. Work outside our respective zones of genius should go to agents wherever possible, and where it cannot, we should build toward making it delegable.
A valid JSON string that a parser reads as salt and a language model reads as an instruction to turn down how much you love your wife.
Most business software is sold like a gym membership. The agentquake roiling the industry is rapidly puncturing the assumptions that made this model work.
The Internet is about to have more agents than humans. AGI is here. Robots will be everywhere. AI agents will operate autonomously. Ok — but what actually comes next? A whole stack of systems, protocols, and infrastructure will need to be built for a world where agent populations exceed human populations.
Vibe coding commodifying engineering has generated fear that all software is now up for grabs. But when execution gets cheap, the definition of what's worth executing changes. Taste and trust become the scarce assets.
Part 2 of our kernel vulnerability analysis. Weekend commits are 8% less likely to introduce vulnerabilities, but take 45% longer to fix. 117 super-reviewers catch bugs 47% faster than average.
What I learned mining 125,000 vulnerabilities from Linux git history. The average kernel bug lives 2.1 years before discovery, but some subsystems are far worse.
From a park bench handshake in 2022 to millions of users creating at the speed of imagination. How KREA is collapsing the distance between ideas and their realization.
We are Pebblebed: builders who invest in builders.