Field notes on operations, AI, and building things that actually ship. Some of these first appeared on LinkedIn.
Agents write the code now. What still needs a human is owning what must be true, and judging how it feels.
Read →Jane is on maternity leave, and Andy has lost someone he loves. Both will be back, and the work has to keep moving until they are.
Read →I built an assembly line of eight AI agents, measured it, and deleted almost all of it. Here is what I'd do instead.
Read →People will quit and people will be laid off. What Bob knew doesn't have to go with him.
Read →How I spent a year teaching machines to remember why, and then noticed that the companies they work for never learned to.
Read →The model improves for your competitor on the same morning it improves for you. The record of what your company decided, and why, is the one curve that belongs to you.
Read →Speed was never the prize. What I had wrong about AI agents for two years, and what Kyroco is here for.
Read →Why language models lose the thread between sessions, and what a real, governed memory layer changes about the way agents work.
Read →What shifts when agents do the building, and how to keep humans on the decisions that actually matter.
Coming soonUsing synthetic audiences to pressure-test product decisions before a line of code gets written.
Coming soon