The Exponential Method
State and ceremony: keep the State explicit, use it to guide work, and return the results to it.
Version 1.1 · 14 November 2026
Keep the State. Close the loop.
An organization makes promises: what matters, what will happen next, who is responsible, and why a decision was made. Those promises need to survive the conversation in which they were made.
When they do not, work begins with reconstruction. Someone reads old messages. Someone asks the founder. Someone discovers that last week’s decision never reached the person doing the work.
Exponential begins with a simple idea: keep the organization’s State explicit, use it to guide work, and return the results to it. People and agents continue from what the organization has already learned and committed to.
Why now
Teams have always run partly on memory. Priorities live in the founder’s head, decisions in a chat thread, status with whoever was asked last. People cope because people can read the room.
Agents can’t. An agent knows what was agreed on Tuesday only if it was written down where it can read it. Once part of your team is AI, unwritten state stops being a quirk and becomes a liability: every agent you add starts blind to how the company actually works.
The fix is not another dashboard. Make the State explicit, keep it current, and change it deliberately, with the people accountable for it.
State is the recorded understanding
The State contains current priorities, work, decisions, commitments, owners, blockers, and open questions. It includes what is uncertain and when information was last checked.
State is not reality. A project may be blocked before anyone records the blocker. A task may be marked complete before its result has been verified. The difference is drift.
The aim is to make important gaps visible and resolve them. A perfectly maintained task list is not enough: the work also needs to move toward worthwhile goals.
The State Loop
The method connects five activities:
- Observe. Capture what happened, with evidence and provenance.
- Interpret. Ask what it means for goals, assumptions, and commitments.
- Decide. Record the choices and responsibilities that follow.
- Act. Do the authorized work.
- Reconcile. Bring the results back into the State and identify what remains unresolved.
These activities overlap. Observations arrive while work is underway. Urgent changes need not wait for a scheduled review. The loop has a rhythm, but it is responsive to events.
Ceremonies make room for judgment
A ceremony is a named, repeatable practice for reading the State, considering what it means, and recording outcomes. It has a purpose, relevant inputs, participants or an individual owner, and an expected form of output.
It can be a daily check-in with yourself. It can be a check-in between you and your assistant. It can be an asynchronous review or a live discussion with a team.
Its agenda begins with what changed, what is blocked, what remains unresolved, and what requires a decision. People can add concerns that the State has not yet captured. The outputs return to the same work: decisions, commitments, corrections, and explicit dispositions.
A ceremony may conclude that the current course remains right. It need not create tasks to prove that it happened.
Principles
1. Make important commitments accessible
An important decision should not depend on finding the person who remembers it. Record enough context for someone joining later to understand what was decided, who is accountable, and why.
2. Preserve continuity across people and agents
The organization’s memory must persist when an assistant is replaced, a colleague is absent, or a team changes shape. Different participants can work through different interfaces while relying on the same accepted commitments, subject to their permissions.
3. Generate a starting agenda; leave room for insight
Use the State to surface blockers, changes, stale assumptions, and unresolved questions. Let participants introduce new evidence and challenge the model. A concern does not need a perfect data structure before it deserves attention.
4. Give decisions a durable home
Record decisions beside the work they shape, with ownership and relevant evidence. Preserve the rationale when a decision is superseded. Tasks can finish while the consequences of a decision continue.
5. Distinguish proposals, authority, and facts
An assistant can propose an interpretation. An authorized person can accept a commitment or delegate authority within a mandate. An integration can report an observed event. These are different kinds of change and should remain recognizable in the State.
6. Verify outcomes at the level of the promise
Approval is permission. An attempt is execution underway or attempted. Completion requires evidence that the intended task was fulfilled. Sending a proposal does not mean the customer accepted it. The State should say precisely what is known.
7. Give unresolved work an explicit disposition
Continue it, defer it, escalate it, cancel it, or archive it. Carry-forward protects continuity; it should not create an immortal backlog. When something is no longer worth doing, say so.
8. Use the smallest useful rhythm
Start with the recurring practice that resolves a real coordination need. Add another when it serves a distinct purpose. Protect time for judgment and work. Respond to urgent events when waiting would be costly.
9. Keep goals open to revision
Relate work to purpose without forcing every obligation, incident, or experiment into an artificial hierarchy. A review should be able to reveal that the goal itself needs changing.
10. Preserve organizational ownership
The organization should be able to understand, export, and continue using its State. Data ownership and deployment control deserve explicit choices. A technical installation claim should make clear what remains dependent on outside services.
A week through the loop
On Monday, a project owner sees that a launch depends on an unresolved integration problem. The evidence enters the State. Tuesday’s review agenda includes the blocker and the decision it creates.
The team considers two options: move the launch or reduce its scope. It chooses a smaller launch, records the rationale, assigns a human owner, and authorizes an assistant to prepare revised customer communication. The communication remains a draft until the required approval.
The assistant prepares it. The owner approves it. Execution records whether it was sent successfully. These are separate events.
At the next review, the team sees the decision, the resulting work, and the evidence. A remaining technical question is explicitly deferred with an owner and a review date. Nobody has to reconstruct the story from last week’s transcript.
Begin with one practice
Choose an existing weekly review or daily personal check-in. Write down its purpose. Identify the current commitments it needs to inspect. End by recording what changed and what remains unresolved. Start the next occurrence from those outcomes.
The method can begin with ordinary documents and a work tracker. Exponential makes the State, its agendas, the outcomes, and the follow-through one connected product.
Foundations
The method draws on feedback loops and OODA’s attention to observation, interpretation, decision, and action. Its state model is informed by explicit state transitions: accepted changes update the State, while execution affects the outside world. Its vocabulary of state, drift and reconciliation comes from infrastructure engineering, where a controller compares the recorded state with what is actually running and closes the gap.