The companies dying right now aren’t dying from lack of AI tools. They’re dying because it takes five meetings to decide if an AI agent should be allowed to answer a customer email.

If you look at the companies that failed in the dot-com bubble, the ones that stalled out in 2008, and the ones flailing today in the face of AI, you’ll notice the same pattern. They treated every decision like a Supreme Court case.

There is a dangerous tendency in growing organizations to invest equal time in all decisions. We schedule a one-hour meeting to discuss a strategic merger, and then we schedule a one-hour meeting to discuss the new travel expense policy.

The orientation to treat all decisions equally is fatal.

Good decision-making is the same in any era. It doesn’t matter if you are running a textile mill in 1890 or a generative AI startup in 2025. The principle remains: You must ruthlessly segregate decisions based on their reversibility.

If you are struggling with speed, it isn’t because you need better tools. Assuming accountability is clear, it’s because your governance is treating reversible experiments like irreversible commitments.

The Two Types of Decisions (And How to Treat Them)

To cure the paralysis, we have to go back to the framework I outlined in Designed to Scale, heavily influenced by the Amazon “One-Way Door” concept. But we need to look at it through the lens of velocity.

Type 1: The One-Way Door (Strategy)

These are decisions that are hard to reverse. If you walk through, you can’t turn around. Mergers, major architectural shifts, major technological investments.

  • Time Allocation: High
  • Who Decides: One individual on the Strategic Execution Team (SET)—your cross-functional decision-making body for Type 1 decisions
  • The Goal: Shared understanding and alignment on why we are making this decision.

Type 2: The Two-Way Door (Operations)

These are reversible. If you make a mistake, you can back up and fix it. Pricing adjustments, marketing A/B tests, internal workflows.

  • Time Allocation: Near Zero
  • Who Decides: Individuals (or AI agents) at the edge
  • The Goal: Velocity and learning

Here is the hard truth: Most organizations are suffocating because they treat Type 2 decisions with Type 1 rigor. They want a committee to approve a reversible action.

In a healthy organization, everyday operating decisions shouldn’t take “time.” They should take courage. You seek perspective (not permission), you decide, you act, you measure the result, and if you were wrong, you decide again.

Why AI Makes This Framework Non-Negotiable

These aren’t new concepts. They’re the foundation of The 3 Agreements framework I’ve used with expansion-stage companies for over a decade. What’s new is that AI makes this framework even more vital.

Without clarity around Type 1 and Type 2 decisions, AI adoption becomes bureaucratic theater. Teams will want committee approval before letting an agent route a support ticket. The SET will waste time debating operational trivia. You’ll have automated nothing but your existing dysfunction.

An AI agent can make thousands of decisions per day—but only if you’ve built the workflows to let it. If your culture requires permission for reversible decisions, your agents will sit idle while humans schedule meetings about whether they should be allowed to act.

The companies that win over the next few years won’t be the ones with the best AI tools. They’ll be the ones with the governance and culture to actually use them.

The 3 Agreements: Proven Framework for High Velocity

If Type 1 and Type 2 thinking is the diagnosis, The 3 Agreements are the cure. I’ve used this framework with hundreds of CEOs to create high-velocity cultures. Here’s how it works in AI-enabled organizations:

Agreement #1: The SET Owns the “One-Way” Doors

Strategic confusion happens when leaders delegate strategy instead of tactics. The Strategic Execution Team (SET) must own Type 1 decisions.

The Rule: Anyone can bring a Type 1 proposal to the SET, but it must be written in full sentences (not bullet points) and accepted by the implementer.

The AI Application: While AI is excellent for crafting proposals and modeling scenarios, the human implementer must ultimately accept the decision and lead the charge.

The Outcome: We slow down to get the big things right and go fast once the decision is made.

Agreement #2: Everyone Seeks Perspective and Accepts Accountability for Their Role(s)

This is where the culture shifts. For reversible decisions, we do not seek consensus. We seek perspective, and then we act.

The Rule: Before making a decision, gather data and seek out perspective both upstream and downstream, as appropriate. Seeking perspective is not the same as seeking permission. If the cost of being wrong is lower than the cost of waiting, you have a green light.

The AI Application: This is where AI transforms everything. Should an AI agent respond to this support ticket? Should this workflow be optimized this way? Should we A/B test this agent prompt? These are Type 2 decisions. The individual (or agent) seeks perspective and acts. If you require permission for these reversible decisions, you’ve just automated bureaucracy, not work.

The Outcome: We stop having meetings about things that can be easily reversed. AI agents execute within agreed-upon guardrails, while humans focus on the strategic Type 1 decisions that shape the future.

Agreement #3: We Commit to Listen, Understand, Provide Perspective, Decide, and Act

Once a decision is made—whether by a human SET member or an operational AI agent—the debate ends, and execution begins.

The Rule: Frank and honest discussion of the facts before a decision is made, followed by total commitment to implementing the solution after the decision is made. “I disagree and commit” is a perfectly acceptable response.

The AI Application: When an AI agent makes a decision within its authority, you let it run. You don’t get to relitigate whether the agent should have that authority during implementation. That’s a Type 1 conversation for the SET. AI agents surface issues and options faster than humans ever could. If you relitigate decisions during implementation, you’ve lost the advantage.

The Reality: In an AI-enabled workflow, answers and issues arise faster than ever. Unless a decision approaches a catastrophic risk threshold, you keep moving and learning.

Structure Supports The Strategy

The bottleneck is rarely technology. The bottleneck is where the throughput bogs down.

If you don’t define which decisions require a meeting and which require a mouse click, you will drift into a state of “permission paralysis.” Your high-performers will get tired of asking for approval to do their jobs, and your AI tools will sit idle because no one knows who is allowed to press “enter.”

Wise governance isn’t about control. It’s about velocity and avoiding truly systemic harm.

It’s about building a structure where the reversible decisions flow like water—fast, fluid, and constant—so that you have the time and energy left to focus on the few, heavy, irreversible decisions that actually determine your future.

Start Here

If you’re serious about implementing this, start with these three actions:

1. Download The 3 Agreements one-pager at organizationalphysics.com/the-3-agreements and bring it to your next SET meeting. Discuss: Are we treating Type 2 decisions with Type 1 rigor? Where are we creating permission paralysis?

2. Look at your calendar for the next week. Which meetings are focused on Type 2 reversible decisions? Cancel them. Delegate to the appropriate person or agent to seek perspective and get it done. That’s Agreement #2 in action.

3. Read my Ten Rules for Highly Effective Meetings at organizationalphysics.com/2024/03/06/ten-rules-for-highly-effective-meetings/. Pay particular attention to Rule #4: defining WHO has authority over a decision before you start the meeting. When you’re clear on the WHO, the rest of the team can orient to helping them make and implement the best decision. This prevents the circular debates and permission paralysis that kill velocity.

Whatever the future holds, the physics of decision-making won’t change. The pace may accelerate, but the principles endure. The only variable is your willingness to apply them.