(And you don’t need to wait for advanced AI before making the necessary changes.)

The Singularity

You don’t need to sit through many AI discussions before someone invokes the singularity — that hypothetical moment when machine intelligence surpasses human intelligence, triggering an uncontrollable explosion of technological progress. Thinkers like Vernor Vinge and Ray Kurzweil made this idea famous.

This concept has now been ported into the world of business. The organizational singularity is the claimed tipping point at which AI collapses coordination costs so dramatically that a single person (or tiny team) can outperform what once required organizations of dozens or hundreds. In its most breathless form, it’s the idea that a 13-year-old in pajamas could launch and scale a billion-dollar business while playing Call of Duty.

The rise of AI has also revived the older idea of the sensing organization — a company designed to continuously detect signals from its environment and respond in real time, rather than relying on rigid make-and-sell planning cycles.

A more sober vision

There is value in these dramatic predictions. They serve as a useful wake-up call. Technology does disrupt, and AI is one of the most consequential technologies in decades.

But there is also danger in the hype. While AI will disrupt virtually every industry to some degree, the nature and intensity of that disruption will vary enormously. The implications for a video game developer are very different from those for an automotive manufacturer, a hospital, or a professional services firm.

I’d like to propose a design framework specifically for traditional businesses — organizations that primarily transact in atoms rather than electrons.

Importantly, this framework does not require AI for adoption. In fact, if you examine many of the fast-growing companies that have disrupted established industries over the past decade or two, you’ll discover they have already been operating according to this model. It was a central driver of their success.

These organizations are now exceptionally well-positioned to harness the true potential of AI. And your organization can be too — if you seize this moment to rethink how it is designed.

Why your organization is not heading for a singularity

To appreciate why your business is not heading for an organizational singularity (assuming you’re not running a pure digital or game development company), it helps to contrast two fundamentally different types of systems.

Consider your body’s insulin response. When you eat a meal, your blood glucose rises. Specialized cells in your pancreas sense this change in real time and release precisely the right amount of insulin — no committee meetings, no annual planning cycle, no batch processing. The response is immediate, proportional, and decentralized. This is nature’s elegant version of a sensing organization.

Now compare that to your home’s HVAC system. Your house has significant thermal inertia — it takes a long time to heat up or cool down. Because of this lag, the thermostat operates with a deadband — an upper and lower limit. If it were tuned as sensitively as the insulin system, the HVAC would chatter constantly, switching on and off in rapid succession.

If your business primarily transacts in atoms (rather than electrons), the HVAC system is the more appropriate analogy. There’s no doubt that AI has transformative potential. But it is highly unlikely that AI will magically transform your business into a frictionless biological system. Your planning, design, and expectations should reflect that reality.

Clockspeed

The HVAC analogy makes one thing clear: faster is not automatically better. But slower isn’t automatically better either.

The real question is one of clockspeed — the natural rhythm at which your organization senses, decides, and acts. Even if we assume infinite intelligence and perfect information, at what speed should decisions be made in an atom-based business? In your business?

The goal of your business is to generate profit. Profit ultimately results from the delivery of value to customers. Therefore, if you want to maximize profitability, your primary objective should be to maximize the rate of value delivery to your customers.

Of course, you should optimize your pricing and costs. But once those are reasonably well tuned, the rate at which you deliver value becomes the single most durable and powerful lever you have for long-term profitability.

Now, here’s a critical insight.

The rate at which your organization delivers value is almost always determined by a single resource (or a set of similar resources). If you can identify this resource, then your organization’s optimal clockspeed is the rate at which this critical resource completes work.

If you want to prepare your business for AI, you should deliberately synchronize your organization’s clockspeed — the rhythm at which it senses, decides, and acts — with the rate at which this resource completes work.

This is harder than it sounds. It requires fundamental changes to how your organization is designed and managed. But, importantly, you do not need AI to make these changes.

If you layer AI onto your current design, you will get some benefit — but it will be limited. However, if you make this clockspeed alignment now, it becomes a major unlock for your business. And when you later introduce AI, it will act as a powerful accelerant rather than an incremental cost-saver.

The clever fix that made clocks run slower

If your business is like most, your actual clockspeed is determined more by the cadence of your accounting department than by the rate at which your primary resource delivers value to customers.

This is not an accident. It is the direct result of a fix we applied to a very real problem many decades ago. The good news is we no longer need this fix. Modern technology is more than sufficient to solve the original problem — yet almost every organization continues to operate as if the problem is intractable and the fix remains essential.

To understand the problem — and the fix — it helps to look at an environment where the problem barely exists. Imagine a small pizza shop, with the owner standing between the counter and the pizza oven. From that vantage point, they can see customers arriving, smell when pizzas are ready, and monitor inventory levels — all at the same time. Predictions are easy. Decisions are fast and accurate. The clockspeed of the business naturally matches the rate of value delivery.

Now contrast that with a larger organization. Here, no one individual has the pizza shop owner’s vantage point. Organizations are complex systems — full of interdependent resources, feedback loops, and time delays. In such environments, it is not merely difficult to predict how a change in one part of the business will impact the whole — it is practically impossible.

At root, every management decision asks the same fundamental question: What impact will this proposed change have on the profitability of the entire organization? Because every management decision is based on a prediction, decision quality degrades markedly as prediction quality collapses.

Given how difficult it is to forecast global outcomes, companies adopted a very clever workaround.

This workaround didn’t eliminate the need for global profit predictions. Instead, it eliminated all the complex calculus associated with making those predictions. Each operator could calculate a synthetic profit number based only on locally observable information — just like the pizza shop owner.

The only caveat was that every calculation had to include an allocation of global costs. That single cost-allocation number carried an enormous burden. It was supposed to represent the entire complexity of the organization — all the internal dependencies, the first-, second-, and third-order consequences of each decision, and the full set of organizational dynamics — all baked into one number.

Cost allocation tables became the practical solution. They enabled managers to calculate synthetic “profit” numbers for products, customers, transactions, projects, branches, and more. This was incredibly liberating.

But this fix — traditional cost accounting — has a serious Achilles’ heel. The profit predictions it enables are only as good as the allocation tables are current. And in most organizations, those tables are updated only quarterly — or even annually.

The result is that your organization’s clockspeed — the rhythm at which it senses, decides, and acts — is limited by the frequency at which these tables are updated.

Your managers may feel like they are making decisions daily or even hourly. But if those decisions are powered by cost-allocation tables that are updated only quarterly, then in reality, the organizational clock clicks only once a quarter.

Only good decisions count

Of course, managers can (and do) make decisions faster than the accounting department updates the cost-allocation tables. But we have to be honest about what that actually means.

Decision quality tends to degrade rapidly as the cost-allocation numbers grow stale. The assumption we make when we talk about an organization’s clockspeed is that those decisions are sound. Bad or poorly informed decisions don’t contribute to the health or growth of the organization. In many cases, they actively destroy value.

This happens so frequently that it has become a source of dark corporate humor. You’ve probably seen examples like these:

• A sales team gets buy-in to a deal based on last quarter’s cost allocations, only to later have to renegotiate prices internally because they daren’t show the customer a price that’s now significantly above market value.
• A division is forced to discontinue a product that was being manufactured on an old machine during operator downtime — until the new quarterly allocation suddenly made it appear “unprofitable”.

The accounting profession is well aware of these issues. They have developed more sophisticated approaches — such as Activity-Based Costing — and are now exploring the use of AI to update allocation tables in near real time.

Deleting the entire cost-allocation apparatus

It might appear that AI is the solution to the clockspeed problem. But this approach is simply a fix applied to a fix.

The better alternative is to discard the entire cost-allocation apparatus altogether.

We already have powerful evidence that this is possible. Look at the companies that disrupted major industries before the current AI wave — Tesla, Uber, Nvidia, Amazon, and SpaceX. These organizations achieved explosive growth by operating at a much faster clockspeed than their traditional competitors.

Another example comes from the airline industry. Long before AI, airlines adopted dynamic pricing. Instead of setting prices based on fully-loaded costs, they began adjusting ticket prices in real time based on remaining seat inventory — their primary value-generating resource.

The speed-based organizational design framework

If we want to learn from these fast-growth organizations, the key question is this: How did they scale from small enterprises into giants without letting the entire organization fall under the spell of traditional cost accounting?

The answer is simpler than most people expect.

A defining attribute of these well-designed organizations is that they clearly identify their primary value-generating resource (their critical resource) and deliberately subordinate all management decisions to the impact those decisions will have on the performance of that resource.

This idea was powerfully championed by Eli Goldratt in his bestselling book, The Goal (1984). In well-managed organizations, profitability is determined by the performance of the Constraint — the one resource that limits the overall rate of value creation. Every other resource should exist to support that critical resource.

In short, optimal organizational design can be summarized as follows. A well-designed organization clearly identifies its primary, value-generating resource, works relentlessly to maximize its productivity, and subordinates every decision across the entire organization to this single objective.

In a typical organization, there is no explicit recognition of a primary, value-generating resource. Instead, the implicit assumption is that value is generated more or less evenly across the entire organization. This leads to the pursuit of productivity improvements everywhere, which in turn creates complex, often chaotic organizational behavior.

This brings us to a deeper insight.

Management decisions must be informed by predictions about how the organization will behave. An optimally designed organization is a complex system, but it exhibits remarkably simple (and therefore predictable) behavior.

A simple solution to the prediction problem

In a well-designed organization, management decisions fall into two clear categories.

First, decisions that directly impact the productivity of the Constraint. The only question that matters is: Will this action make the Constraint more productive or less productive during the period under consideration?

Second, decisions involving supporting resources. For these, no complex prediction is necessary. These resources should have sufficient protective capacity and simply maximize the rate at which they complete tasks as work arrives.

To support these decisions, senior management should provide clear, real-time signals that enable decision-makers throughout the organization to assess the impact of their actions on the Constraint.

While cost accounting will remain useful for financial reporting and major strategic decisions that affect the design of the organization, it is no longer required to support day-to-day management decisions.

The role of technology (including AI)

The proponents of the organizational singularity have an important insight. Modern organizations should operate at a much faster clockspeed than they do today.

But the ideal clockspeed is not the speed of electrons. It is the rate at which your organization’s primary value-generating resource (its Constraint) can create value for customers.

All technology — including AI — should be deployed with one primary question in mind: How does this technology improve the performance of our Constraint?

A local cost reduction is nice. Automating a supporting activity to make it cheaper or faster provides an incremental benefit. But your return on technology investments is likely to be at least an order of magnitude greater when it is deployed to directly increase the rate at which your Constraint generates value.

The organizational design framework described here — what I call the Speed-Based Operating System — creates a kind of singularity of its own: the tight synchronization of management decisions with actual value creation. It does not require advanced technology to implement, but it creates the optimal environment in which to deploy new technology — especially artificial intelligence.

More on the Speed-Based Operating System.