What I do

Lead the change with experiments

A new method does not spread by document. One team proves it, with a date to kill it or scale it, and then teaches the others.

The problem

A new method has no value until an organization runs on it, and getting there is its own piece of work. Rolled out by slide deck and mandate, it does not spread. People change how they work when they see a method producing results on real work; a document or a slide deck rarely changes anything.

How I approach it

  • Experiments with a date. One team proves something, with a date set up front to kill it or scale it, and then teaches the others.
  • Adoption you can see. People adopt a new way of working when they see it producing results on real work. So I start with one team using it on their actual projects, where the rest of the organization can watch what it delivers. I count it as adopted when most of the organization works that way, not when a few enthusiasts do.
  • Decision rights written down. The organization writes down who decides scope, who decides the date and who decides priority. Whoever is held to a fixed date can trim the scope to meet it, as long as every commitment already made to a customer is still met. A change that would break a customer commitment goes to whoever owns that commitment.
  • Alignment over agreement. People do not have to agree with a decision to carry it out together. When someone disagrees, the disagreement is written down as a risk, so it is tracked instead of reopening the decision.
  • Decide instead of waiting. Not making a decision is worse than making an imperfect one, because while nothing is decided nobody can move.
  • Good enough is good. A model that works most of the time is a good model. I do not hold a decision back waiting for one that covers every case.
  • Protected capacity. The people buried in the old work are too busy doing what an AI could already do to go build the thing that would end it, so I carve out capacity for it on purpose.
  • A failed idea returns only for a reason. An idea that failed before is not launched again under a new name. It comes back when something has changed enough to make it work now, such as new tools, a new team or a different market, and the relaunch says what changed.
  • Different thinking where the problem needs it. I push back on incremental goals when the problem needs different thinking, and I do it without swinging the organization back and forth between directions.
  • It outlives any one person. The capability has to keep working when any one person leaves or moves on.

What it produces

The change running in the organization’s own teams, owned by its own people.

In the Luci House

The AI-staffed team I run every day as a working prototype.

The house is the experiment. It has run every day since June 1, 2026 (172 days), and I write up what it does and what it teaches in the Lab as it goes, with pictures of it working. New AI members first work in the Playground, the place in the house where I try them out before they join a team. A unit like the house could be running in a real organization, and it can be prototyped there, with one team proving it first. The Luci House, in the Lab

Figures as of November 20, 2026.

All of What I do