By Erno Marius Obogeanu-Hempel
As a long-time OKR expert, I have observed over the past few years how the pace of technological change has shifted dramatically. Moore’s Law was long the gold standard, but today the exponential growth of AI technology far exceeds those cycles. This raises an uncomfortable question for companies: Does the classic 3-month OKR cycle still align with the pace at which innovation is happening today?
Let me start by saying: not everywhere anymore. And that’s exactly where the opportunity lies.
Until now, Moore’s Law has held true
Moore’s Law describes the observation made by Intel co-founder Gordon Moore that the number of transistors on a microchip—and thus the available computing power—doubled approximately every 18 to 24 months over many decades. Although this trend has since slowed due to physical limits, Moore’s Law remains a symbol of exponential technological progress to this day. At the same time, new sources of innovation are emerging, such as specialized AI chips, software optimizations, and more powerful algorithms (see OECD: “The End of Moore’s Law? Innovation in Computer Systems Continues at a High Pace”).
AI: Moore’s Law on Steroids
Researchers are currently observing a similar exponential trend in the performance of modern AI systems. The research organization METR (Model Evaluation & Threat Research) does not measure the general intelligence of AI, but rather how long an AI agent can successfully handle complex tasks on its own. Over the past few years, this so-called “task horizon” has doubled on average about every seven months. More recent analyses even suggest a further acceleration: Since 2023, the doubling time has been around 4.3 months, and since 2024, it has in some cases been as short as about 2.9 months. These figures refer exclusively to the measurement method used by METR and should not be interpreted as a general doubling rate for overall AI performance (Source: METR, “Measuring AI Ability to Complete Long Tasks”).
Keep this number handy next to your OKR calendar: A doubling time of less than three months is shorter than a single standard OKR cycle.
AI as an Innovation Engine: The Boost Is Felt Everywhere
The implications of this development extend far beyond AI itself. Many scientists view artificial intelligence as a general-purpose technology—that is, a foundational technology that not only advances on its own but also accelerates innovation in numerous other fields of technology. The OECD even describes AI as the “invention of a method of invention”: an invention of a method for inventing.
AI is already supporting research and development in fields such as medicine, materials science, robotics, chip design, and software development, significantly shortening development and innovation cycles in these areas. It automates and scales complex optimization, simulation, and design processes.
The combination of low-code and AI-powered tools also shifts where value is created: Product managers can build functional MVPs on their own without in-depth programming knowledge, while developers focus on complex architecture and true innovation. Time-to-market and iteration cycles are shrinking—not just in software development, but across entire business models.
The OECD addresses this development in its report *Exploring Possible AI Trajectories Through 2030*. One of the scenarios discussed there, “Progress Accelerates,” describes a possible further acceleration of technological progress driven by AI. Important: This is explicitly a model scenario, not a forecast.
Problem 1: The 3-month cycle conflicts with the pace of innovation
What does this mean for OKR? The standard cycle is three months; some companies use a four-month cycle. This cadence dates back to a time when a quarter was a meaningful timeframe for learning and delivering results.
But if relevant developments occur in your field within a matter of weeks, a structural problem arises: The goals you set in Week 1 may already be outdated by Week 8. At that point, you’re no longer steering—you’re merely documenting. And if the competition accelerates its innovation cycles while you stick to your usual pace, you’ll experience a measurable loss in adaptability, market feedback, and time-to-market.
Problem 2: Not all departments work at the same pace
The second problem is more subtle and is often overlooked in practice: Acceleration does not affect a company uniformly. Product development, marketing, and data teams are experiencing real productivity leaps thanks to AI tools and can now deliver in weeks what used to take a quarter. Other areas—such as those dealing with regulatory frameworks, long procurement cycles, or stable operational tasks—simply have no reason to double their pace.
A uniform cycle for everyone then forces either the fast ones to wait or the steady ones to rush unnecessarily. Both cost focus.
Approach 1: Shorter cycles, for example, two months
The first approach is the obvious one: shorten the cycle. If teams can deliver reliable progress in two months and the market or technology demands it, a two-month cycle is not a radical reform but an evolutionary adjustment. The key criterion is this: the duration of the iteration must correspond to the actual speed of implementation.
Shorter cycles mean faster feedback loops, earlier course corrections, and goals that are still relevant at the time they are evaluated. But they also mean more planning, more reviews, and more coordination. More on that in a moment.
Approach 2: Differentiated Cycles with a Shared Heartbeat
The second approach is more challenging, but for many organizations it is the more appropriate one: an overlapping rhythm. Dynamic units operate on a two-month cycle, while more stable units operate on a four-month cycle. The four-month cycle corresponds exactly to two short cycles, so that both rhythms synchronize regularly.
The analogy I use for this in consulting projects is: The company-wide heartbeat remains in sync, but not every organ beats at the same rhythm. For this to work, three things are needed:
- A common synchronization point. At every point where the two cycles overlap, a cross-cycle alignment takes place: strategy review, dependencies, and resource conflicts.
- Stable, overarching goals. Company-wide or divisional OKRs set the direction over the longer term. The short cycles visibly contribute to this, ensuring that each unit’s contribution remains clear, even if the pace varies.
- Transparency regarding the clock logic. Every unit needs to know why it operates at its own pace. Otherwise, differentiation can quickly come across as unequal treatment.
The success of this model hinges entirely on the people who facilitate it: OKR Coaches, Masters, and Champions who orchestrate planning across different cycles and identify dependencies early on. It is precisely this role that we train participants to fulfill in the OKR Coach, Master & Champion training program, including how AI tools can lighten the load of cycle-based work.
Is that even manageable? This is where AI comes back into play
The valid objection to 2-month cycles is that overhead increases. Six planning sessions per year instead of four, more frequent reviews and retrospectives, and more communication effort. This is exactly where the circle closes, because the same technology that speeds up the cycle is what makes that faster cycle manageable in the first place.
Specifically, AI today provides support throughout the entire OKR cycle: in preparing planning sessions (data analysis, hypotheses, draft goals), in formulating and refining Objectives and Key Results, in analyzing ongoing progress based on existing system data, and in preparing reviews and retrospectives. What used to take several workdays of coordination per cycle can now be significantly streamlined.
Based on this conviction, I’ve developed prompts and AI tools that help OKR coaches do exactly that: prepare planning sessions more efficiently, review and optimize goal formulations, and navigate cycles with less friction. Not because AI is meant to take over goal-setting, but because it reduces the process overhead to such an extent that shorter cycles become possible without exhausting the organization. In the seminar “AI for OKR Coaches, Masters & Champions,” I teach how AI can be used specifically for OKR work.
The rule of thumb I share with leadership teams: Don’t shorten the cycle until you’ve reduced the process overhead per cycle. Otherwise, you’re trading speed for overload.
Conclusion: Adaptability is the new discipline
OKR was never intended to be a rigid ritual, but rather an operating system for focused goal-setting. When the environment is accelerating exponentially, the rhythm of this operating system needs to be reevaluated: shorter cycles where the speed of implementation allows, longer cycles where stability is the right answer, and a shared heartbeat that strategically ties both together.
Companies that take this step benefit in two ways: They set goals that align with their teams’ actual pace, and they build the ability to consciously control their own rhythm. It is precisely this ability—not a specific cycle format—that determines strategic sustainability in the age of AI.
Sources: OECD, “The End of Moore’s Law? Innovation in Computer Systems Continues at a High Pace” · METR, “Measuring AI Ability to Complete Long Tasks” (metr.org) · OECD Artificial Intelligence Papers: “What Technologies are at the Core of AI?” / “Is Generative AI a General-Purpose Technology?” · OECD, Exploring Possible AI Trajectories Through 2030
