AI & Agents
Models, agents, and how AI products actually get built — separate from the infrastructure that runs them.
The Difference Between Nothing and Zero
In the mid-1970s, GEICO’s stock fell to around two dollars a share. The company had grown by pricing risk too cheaply, which works beautifully right up until the claims come due — and then the claims came due. Most people looked at that chart and saw an ending. One investor looked at the same chart…
A Lead Is Not a Place You Can Lock
This week a restriction came off — quietly, the way these things tend to move now. The federal office that governs what advanced technology can cross a border lifted a hold that had frozen a frontier AI lab’s most capable models in place. For a while, the best versions couldn’t travel. Now they can. No…
The Distance Between Looking Ready and Being Ready
In the mid-1970s, GEICO was a few months from running out of money. The stock had fallen from around sixty dollars to about two. The company had spent years writing more policies than it could afford to pay out, which is a pleasant way to grow right up until the bills arrive. A new chief…
The To-Do List Is the Last Thing AI Wants
Three million people found OpenClaw in a matter of weeks. That number is the tell. Not because the product was polished — it wasn’t. It required a command line, a willingness to read instructions, and a certain tolerance for things going sideways in the background while you slept. Most people don’t have that tolerance. But…
The To-Do List Is Just the Beginning
Three companies announced personal AI agents this week, all promising to handle your calendar, your emails, your shopping, your errands. The framing is always the same: finally, an assistant that does the boring stuff so you can focus on what matters. It’s a reasonable pitch. It’s also the wrong level of analysis. The to-do list…
When the Interface Disappears, What Remains?
A free tool called OpenClaw reached 3.2 million users in weeks. It let anyone send a message on WhatsApp, go to sleep, and wake up to find an errand done — meeting booked, email drafted, something purchased. It spread that fast because it solved a real thing: the friction between wanting something done and having…