startups
The Room Bill Campbell Insisted On
A management book making the rounds in engineering circles has one blunt instruction buried in its chapter on one-on-ones: shut off your computer, use a notepad, and never take the meeting in a car. It reads like etiquette. It is actually a claim about signal and noise. Most of what a person needs to tell…
The Mundane Edge
A founder at a demo day last spring did not show what her AI model could do. She opened with a customer’s bug report, scrolled through each line item, and explained how the latest release had addressed every one. The investors who had been checking their phones leaned forward by the third screen. The bug…
The Stories That Outrun the Facts
In June 1914, Commissioner Young, Adegboyega Edun, and J. O. George walked to the border between Egba and Iperu to settle a boundary dispute. They carried papers and authority, backed by the quiet conviction that lines on a map could settle what centuries of shared use and occasional violence had not. Two more officials joined…
The $250 Million Bet on Staying Narrow
A valuation near $250 million buys control of Puck this week. RedBird Capital is the buyer, the existing institutional investors are the sellers, and the company changing hands is a five-year-old digital outfit with a tiny staff and a roster of named journalists who cover four rooms: Silicon Valley, Hollywood, Washington, and Wall Street. Nothing…
The Instrumentation Trap
Measuring a thing precisely and controlling it are different skills, and the gap between them is where most startups actually die. That gap has been widening for a decade, because instrumentation keeps improving and the willingness to be corrected by it does not come bundled with the software. You can chart a storm in perfect…
Count What You Have Before You Build More
A $400 million deal closed this week that says more about the AI buildout than any model release did. The first firms that made their money financing GPUs, the physical shovels of the training boom, are moving that capital into inference chips. Not the hardware that teaches a model. The hardware that runs it, over…