I’m excited to share this guest post by Phil Le-Brun and Jana Werner, co-authors of The Octopus Organization: A Guide to Thriving in a World of Continuous Transformation. I know you will learn a lot from their insights into how to drive lasting change and a nimbler organization in the face of the kind of continuous change we face in today’s landscape, technological and otherwise. Enjoy!!

Recently, in two separate conversations, two Fortune 500 CEOs from different industries shared the exact same frustration with us about their efforts to leverage AI: their teams weren’t “imaginative.”

We challenged them. The issue wasn’t a lack of imagination. We saw sparks of brilliant innovation everywhere. The real culprit was their organizations’ operating systems smothering the ideas. This is in itself not unusual for large companies, but it’s a tendency that has been enhanced by the inability of these slow corporate machines to adapt to two exciting forces that were introduced by AI. At the core of the problem lies the addiction of traditional systems to certainty before action.

Two powerful new AI forces

First, AI capabilities are advancing at a pace unlike anything we’ve seen in the history of technology. Yet prevailing top-down decision models assume the centre can absorb new information, decide what “good” looks like, and hand answers down to the edge. That setup no longer works.” By the time the centre has decided, the ground has moved. The people closest to the edge are often the best placed to discover how this technology can be applied to create value.

Second, the cost of delivery has collapsed. The effort and capital required to build a prototype and test an idea have dropped to near zero. . Yet most of our governance was built for an era when delivery was expensive and slow and it was worth spending months justifying an idea before pursuing it. Today, the economics have reversed. When creation is cheap, you need to experiment to discover what’s worth doing.At Amazon, when we need to decide between two competing prototypes, we now simply build both.

Imaginative and creative people are the most valuable type of talent for companies who want to match the speed of AI. If these people have to assemble resources across silos, draft elaborate proposals for grand committees, and spend political capital just to run a simple experiment, they lose precious energy that should be spent on innovation. That’s the human cost of certainty addiction. And we’ve experienced countless examples of this in our roles as Executives in Residence at Amazon Web Services, when we advise executives and transformation leaders of Fortune 500 and public sector organizations.

When Permission Takes Longer Than Execution

Consider a financial services engineering team we watched use advanced tools to modernize 100,000 lines of legacy code over a single weekend, saving three months of manual effort. Yet their security review board insisted on a manual, line-by-line audit of the new code, a legacy process scheduled to take four months. The execution took forty-eight hours. The permission took a third of a year.

Similarly, a global consumer brand wanted to localize a new product launch campaign across 24 countries. Using modern digital tools, they translated the scripts, adapted the imagery for the local cultures, and generated localized video ads for every country over a single lunch break. The company’s traditional Global Brand Integrity Board then stepped in, requiring manual, market-by-market compliance checks by local agencies to approve the tone. The creative execution took two hours. Getting approval took three months.

Somewhere along the way, we confused careful decision-making with guaranteed outcomes. That logic may have been rocky before, but it spells disaster in a world shaped by growing uncertainty and volatility.

That’s why we were so excited to be invited to feature on Annie Duke’s blog. Annie has spent her career making one truth visible: every decision is a bet. You weigh what you know, what you don’t, and you place a wager on the future. The outcome depends partly on the quality of your reasoning and partly on what the world happens to do next. The corporate operating systems we inherited were built to deny this dual dynamic. They treated decisions as deterministic acts: do the analysis, follow the answer, expect the outcome. That worked when the world was slow enough to pretend. It doesn’t now.

The Cost Of “Resulting”

Once you see decisions as bets, the behavioural engine of artificial certainty becomes obvious. Annie has named it “resulting”, the tendency to judge the quality of a decision by how it turned out. A sound, well-reasoned bet that lands badly gets graded as a bad decision. A reckless bet that happens to pay off gets graded as a good one. Outcomes are part decision, part luck. Resulting confuses the two.

In corporate hierarchies, “resulting” runs one way. Bad outcomes are punished even when the underlying logic was sound. Good outcomes from sloppy reasoning get celebrated. Employees pick up the asymmetry quickly, and they draw the rational conclusion: never be the person standing next to a bad outcome.

That fear is what produces the hesitation we see in the boardroom. If any bad outcome can end your career regardless of how good the bet was, the safest move is to never place a bet you might be held responsible for. The result is paralysis, learned helplessness deflecting: decisions are deferred through analysis and advice seeking, while responsibility is pushed further up the chain.

Experiments are by nature uncertain. Perceiving them as a straight line to a business outcome that is A. wrong or B. right, depending on the decision was always faulty. To their core, they are and have always been information drivers. But now that this information can be obtained cheaply and quickly, the cost of inaction has become enormous.

Trust the Arms

It brings us back to our two frustrated Fortune 500 CEOs. Their framing of the issue as a lack of imagination of their teams misses the real problem. Their teams don’t lack imagination; they are just brilliantly adapted to survive an operating system that rewards compliance and blame avoidance over making a bet.

So what, then, should they and anyone else bumping into this type of conundrum do? Well, to put it in the wise words of Peter Drucker: “what gets measured gets managed”. The backlog of stalled decisions inside your company is almost never tracked. It should be the first number your autonomous agent reports every morning. Once we have this information, the instinctual move might be to issue a mandate telling your team to “be bolder.” But we cannot exhort our way out of a systemic design flaw. If we leave the electric fence of “resulting” turned on, our people will continue to choose the safety of central deference every time.

The fix is not to push them to make better calls. It is to adapt an organization where placing good bets is the path of least resistance. To do that, we need a different operating system, one built on distributed intelligence. Nature solved the same bottleneck millions of years ago, without funneling every signal through one central decision-maker. As we described in our book The Octopus Organization, this wildly intelligent animal doesn’t route every sensory input back to a central brain. Instead, two-thirds of its neurons live directly in its arms, empowering them to sense, react, and make autonomous decisions at the perimeter. Organizations need the same biological shift. The lesson is this: you don’t need to find more imaginative people. You need to design an environment where the “arms” of your organization are trusted and equipped to act on what they see and imagine.

The Freedom to Bet

Here are five practical ways to trade artificial certainty for the freedom to bet and imagination to flourish:

1. Default to the two-way door

Most decisions inside organizations are reversible, or what at Amazon we call two-way doors. You can walk through, see what happens, and walk back. A handful are not. The trap most companies fall into is treating every decision as a one-way door, applying the slow, careful machinery built for irreversible commitments to choices that could be safely undone in an afternoon. Annie even argues most decisions are two-way doors except death, you just have different degrees of risk associated with them.

At Amazon, we make this distinction explicit, and we ask leaders to act on roughly 70% of the information they wish they had for two-way doors and treat action itself as the next data point. They are encouraged to reserve heavy analysis and senior sign-off for the genuinely irreversible: large capital commitments, public commitments, decisions that touch customer trust or safety. Everywhere else, the question is not “are we sure?” but “is it cheap to find out?”

2. Fund the prototype, not the PowerPoint

When the cost of execution drops toward zero, the relative cost of deliberation goes through the roof. In the legacy world, we forced teams to spend months constructing elaborate, speculative slide decks and run the approver gauntlet of “no” just to get permission to test an idea. It was an exercise in creative writing, not risk management.

AI flips that corporate math. If it is faster and cheaper to develop a prototype than to schedule a steering committee, stop talking and start building. When faced with two competing ideas, do not debate them. Test both. Let real user data, not the loudest voice in the conference room, tell you which one wins. Organizations that thrive going forward are able to quickly find out what works and what doesn’t.

3. Pre-commit to the off-ramp

The reason small bets feel scary in large organizations is rarely the cost of placing them. It is the fear of being the person who has to admit one didn’t pay off. Sunk costs and commitment bias make us hold on to our experiments, so does loss aversion. Once we invest a little time, we hate to let them go. So bets either don’t get placed, or they turn into zombie initiatives that no one will quit. This is both an issue when traditional organizations commit to few large initiatives, and also when new organizations run a proliferation of cheap fast AI-supported prototypes and proofs of concept.

Annie’s antidote, drawn from her book Quit, is “the kill criterion”. Before the bet is placed, define the explicit, objective signals that will require the project to fold. Spell out exactly what failure looks like, whether it is a lack of user adoption, an unviable cycle time, or a missing customer signal, before the emotional fog of execution rolls in. Pre-committing to the off-ramp makes the bet safer to place, because everyone knows in advance how it ends if it doesn’t work. If the bet crosses that threshold, kill it with pride and pocket the data.

4. Replace gatekeepers with guardrails

Velocity requires a shift from top-down permission to decentralized ownership. If every micro-experiment requires the approval of a centralized approval committee, organizational learning grinds to a halt. The line-by-line code audit and the market-by-market brand check are not unusual stories. They are how most large organizations still operate.

Leadership’s job in the AI era is not to approve each step. It is to design the perimeter inside which teams can move freely. Establish the financial, legal, security and brand boundaries up front. Encode them into the tools and platforms where the work happens. As long as a team’s bet falls inside those boundaries, no signature is required.

5. Grade the decision quality, not the outcome

In a world of growing uncertainty and volatility, you cannot grade your teams only on whether the bet paid off. A smart experiment can land badly by nature. If you punish smart decisions that turn up tails, you teach people to stop running them.

Change your review ritual. Retire the question “did it work?” Ask three different ones. Was the hypothesis logical given what we knew? Did we test it safely and cheaply? What did we learn that shifts the odds on the next bet? Document the logic of the bet before the result is known, so the decision can be evaluated separately from how it landed. A failed small bet that yields deep customer insight is a win for the organization.

The frustration shared by those two Fortune 500 CEOs is real and almost universal. It’s a valid concern, but the framing is wrong: their teams don’t lack imagination, they lack permission. Their people have developed a rational instinct for survival in an operating system trapped by centralized control. The latent capacity to imagine and build is already there, waiting for an environment that lets it move. Unlocking imagination isn’t a matter of pushing teams harder. It is a matter of reshaping our systems for a world where technology advances at a pace we have never seen before and the cost of delivery drops toward zero.

The tools are ready. The imagination is there. The only question left is whether you are willing to let go of the illusion of certainty and trust the arms.

You can find more of our ideas on our Substack: The Ink Tank https://substack.com/@octopusorg/ or at www.theoctopusorganization.com

See you there,

Jana and Phil