Green Is a Measurement, Not a Decision
Nearly a quarter of what goes wrong is a check that said yes. Even when the check is right, green is still not a ship decision.
Technology executive and hands-on architect with 11+ years building and scaling engineering organizations. Three-time company builder with a successful acquisition under the belt.
Technology executive and hands-on architect with 11+ years building and scaling engineering organizations and shipping products to hundreds of thousands of users.
Three-time company builder: grew a 30-person engineering org at AWS, co-founded and sold a blockchain studio via acquisition by Dapper Labs, and took a healthtech startup from napkin sketch to 50K+ users as founding CTO.
Currently Chief Architect at Tarobase (poof.new), building AI-powered tools for vibe-coded dApps. Proven ability to set technical strategy, recruit world-class teams, and raise venture capital.
Nearly a quarter of what goes wrong is a check that said yes. Even when the check is right, green is still not a ship decision.
Gartner expects more than 40% of agentic AI projects to be canceled by the end of 2027. None of the three reasons it names is a model-capability problem. The gap that kills agents in production is organizational, and we already know how to close it.
The most-cited number about AI and jobs is real, and almost everyone repeats it wrong. Here is what the Stanford paper actually found, what its authors refuse to claim, and the one fact in it worth building a hiring plan on.
Nearly a quarter of what goes wrong in multi-agent systems is a verification failure. The most common one is a check that ran, and said yes.
You debug the agent that produced the bad output. The bug was in the message it got handed.
Always open to interesting conversations and building things that matter. Based in Washington, DC.