Field Notes.
Short observations from the three dimensions — written from the ground, not the brochure.
One practice, three principals, three dimensions. Each note is written by the person who works that dimension — Human, Organization, Technology — and points back to the framework it draws from. The full method lives in the H.O.T. Framework.
The framework is published and open. Our work is to bring it to more organizations — run across all three dimensions at once, so the human and technology legs are there from the start.
The structure has changed. The humans haven’t.
Same tools, same budget, different results. What decides whether an AI rollout holds is the person watching the agent do their job — and whether anyone did the human work before it went live. Article 2 of the series.
Becoming AI-native is not a technology problem. It is a readiness problem.
Almost everyone is using AI. Almost no one is getting paid for it. The organizations that stall didn’t have worse tools — they had people and structures that were never ready to use them.
The matrix sorts the tasks. Nobody asks what held them together.
A task can score a 5 for agent-readiness and still sit inside a strong bundle. Two columns the Task Decomposition Matrix needs: where does the judgment go, and what does separating this cost?
The scarce thing moved.
When intelligence becomes abundant, value relocates to whatever is still scarce — judgment, taste, conviction. The dimension everyone underfunded just became the differentiator.
Your successful pilot is why nothing shipped.
A failed pilot forces a decision. A successful one lets everyone feel the question’s been answered — while the authority that would actually move it stays exactly where it was.
The destination is Salim's. The discipline of arriving is the practice.
ExO 3.0 is the methodology for the hardest of the three dimensions — the deepest, most operational leg. H.O.T. completes it with Human, Technology, and one property the Intelligence Stack doesn't name on its own: ownership.
The agent proposed. The store said no. Then a person decided.
An agent shopped a store with a spending limit. The store refused the cart that went over, and nothing moved until a person raised the limit. The wall can refuse; somebody has to own the yes.
Diamandis drew the map. Two coordinates came back blank.
The Death of the Company carries Ismail's Organizational Singularity to its widest audience yet. Two of the three things Diamandis says survive — human judgment and owned intelligence — arrive underspecified. This is the org discipline that holds them.
You find out what a system can do by watching it do something you didn't plan for.
An agent called its creator at 3am; a model emailed from inside a sealed sandbox. Capability is discovered, not specified — so the controls that survive are the ones that don't depend on predicting what the agent will do.
The workshop everyone loved, and the Monday nothing changed.
Enthusiasm in the room is the feeling most often mistaken for change. Monday is the real test — and by Monday the wiring is exactly where it was.
A better agent doesn't need less governance. It needs more.
The intuition that better models need looser guardrails is exactly backwards. Capability and consequence scale together — so the wall matters more, not less.
Readiness comes before building. Where do you stand?
The method's spine is step 2 — score honestly before you build. The H.O.T. scan is that read across all three dimensions: Human, Organization, Technology. Free, and yours to keep.
One scan, one score per dimension, kept in your free workspace.
Run the H.O.T. scan → HUMAN · ORGANIZATION · TECHNOLOGY · ALL THREE IN ONE SCAN · IN YOUR FREE WORKSPACE