Co-authored with Steve Wunker and originally published in the journal Leader to Leader.
The great leadership challenge of the artificial intelligence (AI) era isn’t how to integrate a new form of technology. It’s how to reimagine the very nature of leadership itself. As AI moves from novelty to necessity, organizations face a sobering truth: traditional structures can’t keep up.
Most enterprises still operate with 20th-century hierarchies optimized for consistency and control. Information flows upward for approval; decisions are pushed back down for execution. Often, “digital transformations” that promise fundamental process change leave this hierarchy untouched.
In the effort to make organizations faster and nimbler, we have flattened them but often failed to evolve them. Too many decisions still get bottlenecked by senior leaders who insist that processes must flow through them. Twenty years ago, business leaders talked a lot about the need to close the gap between strategy and execution. Then, the idea was to perform
strategy and execution simultaneously (agile and continuous delivery). Today, change comes so quickly that execution often occurs before a new strategy is even formulated. The result is an unhealthy rhythm: agile teams sprint ahead, then headquarters slams on the brakes while it attempts to adapt retroactively. Momentum stalls in a spasm of organizational arrhythmia. A new approach is needed.
AI changes the mechanics of how organizations work.
AI changes the mechanics of how organizations work. By synthesizing information, generating insights, and acting at speed, AI removes many of the reasons that led to centralized command in the first place. This moment calls not for tweaks, but for transformation – a new archetype for leadership.
That archetype is the Octopus Organization.
Why an Octopus?
As innovation practitioners, we’ve spent our careers guiding our own and our clients’ teams through periods of disruptive change, helping them develop new products and frontline technologies, identifying major opportunities, and growing rapidly into new markets—essentially, helping them become disruptors themselves.
For instance, Stephen is the Managing Director of New Markets Advisors and has advised dozens of companies on AI usage. He was a longtime collaborator with the late Clayton Christensen, Harvard Business School’s legendary scholar of business disruption. Stephen played a key role in refining and applying Christensen’s theories of Disruptive Innovation and Jobs to be Done. Jonathan is the Futurist-in-Residence at Amazon, Executive Chairman of the Center for Radical Change, and former Global Futurist and Research Director at HP.
The Octopus Organization framework is based on our work as pioneers and doers as well as in-depth discussions with more than fifty leaders in AI, academia, and industry. We studied dozens of organizations that are moving concertedly to become AI-native, and assessed over two million workforce surveys conducted with the Harrison Assessment team.
The octopus emerged as a helpful metaphor for an AI-native organization that leans into the technology’s strengths in democratizing skills and information. The octopus may seem almost alien-like to us. It has no bones, no central nervous system as we understand it, and nine brains—one in its head and one in each arm. Each arm is capable of acting independently, and yet the animal functions as a coordinated whole. It is a living embodiment of distributed intelligence.
The analogy is clear. In the AI Age, we need organizations that are fluid, ever-sensing, fast-moving, and intelligent at the edges. We need decision-making that doesn’t climb ladders but emerges locally, informed by real-time data and guided by organizational intent. To lead such an organization demands a shift in how authority is structured, how culture is cultivated, and how leaders at every level view their role.
How AI Changes What’s Possible
AI can provide unprecedented contextual awareness, fine-grained decision support, and clear networks of communication at scale. Large language models allow even junior managers to see the wider chessboard that was once the sole province of senior analysts. These systems, when based on appropriate data, bring the right information to the right people at the right time, making context-rich decisions possible across the organization.
Agentic AI, autonomous AI systems that can plan, reason, and then complete complex, multi-step tasks, curate data to suit the needs of managers across the firm and ecosystem. They digest the information accurately and spotlight its most critical implications. As managers formulate their responses, AI flags their biases, tests scenarios, and recommends guardrails, all
in real time. Software improves executive judgment at every level, allowing even junior staff to make complex and risky decisions with confidence. Interfaces and agentic frameworks allow the organization’s “arms” to trade information laterally, instead of feeding it up or down. Like employees, leaders can also access real-time insight about what is going on across the organization, providing the confidence to remain hands-off.
With AI-supported command, control, and communication, strategy no longer chases execution; both are one.
Devolving Leadership
In an Octopus Organization, power no longer resides exclusively at the top. It is thoughtfully distributed to the edges.
That doesn’t mean chaos. On the contrary, it means coherence without direct control. Leaders must define the risk bands – boundaries of decision-making authority based on the impact and likelihood of potential risks—within which teams can operate with confidence. They must invest in the systems, structures, and skills that allow distributed decision-making to thrive. The goal is responsiveness, creativity, and speed.
The goal is responsiveness, creativity, and speed.
At the front lines, this means giving people the tools and authority to act. In a traditional insurance firm, for example, a junior underwriter might spend years learning what an experienced professional knows. Today, with AI-enhanced systems that ingest aerial imagery and risk data, underwriters can make smart decisions far earlier in their careers.
At the insurance giant Travelers, for instance, domain-trained large language models are enabling claims professionals to access specialized knowledge instantly. This frees them from chasing information and allows them to focus on judgment and service.
As you provide freedom, remember an irony: it is often boundaries that allow freedom to flourish. Consider research on how children use playgrounds. When children’s play was assessed on a wide-open playground, it was clear that they stuck closely to the big equipment at the center. The picture was different at playgrounds with a fence. There, the children felt safe to use the whole facility, and they roamed freely right up to the edge.
It is often boundaries that allow freedom to flourish.
In an organization, the “playground fence” may be much further out than frontline employees assume it to be. Make the lines clear so that staff can explore the full space available.
With these changes, frontline work can be both more challenging and more fulfilling. Look at Stripe, a fintech company that is revolutionizing the way businesses accept payments, manage revenue, and operate globally. In March 2025, Stripe released its Optimized Checkout Suite, an AI-powered solution that dynamically adjusts payment method ordering and handles fraud intervention. Based on Stripe’s extensive payment datasets ($1.4 trillion in annual payment volume), the Optimized Checkout Suite can determine the most relevant payment methods to display based on customer attributes and purchase details, leading to an average 12% increase in revenue and a 7% increase in conversion rates. The system also dynamically adjusts checkout interventions based on the likelihood of different types of risk. This reduces fraud rates by 30% with minimal impacts on conversion.
The system helps customers, but it also removes a category of low-impact, low-skill tasks from Stripe’s risk team, allowing them to focus on more nebulous cases. AI increases the volume and complexity of their average workload, but it provides the tools that allow team members to tackle it effectively: a virtuous cycle. These changes at the edge have ripple effects across the organization.
Interested in learning more about the Octopus Organization?
Redefining the Role of Middle Management
Middle managers, long viewed as custodians of process and approval, face a turning point. As AI takes on the mechanics of information processing and reporting, these leaders must evolve into coaches, connectors, and capability-builders.
Their role becomes less about directing and more about enabling—asking the right questions of AI systems, guiding teams in judgment, and surfacing risks that machines might miss. This is a significant cultural shift. It requires training, time, and trust.
Look at what happens now at HelloFresh, the world’s largest meal kit delivery company. Historically, middle managers in operations spent much of their time planning production runs. But the company has now embraced hyper-personalized meal selections (something enabled by AI-fueled analysis of customers’ previous choices) and that means the kitchens are producing an almost infinite variety of dishes. Creating an Excel model for all that would be impossible, so AI now sets the production plans. Middle managers critically evaluate if the plans make sense and whether the AI model has taken into account all relevant factors in the facility. Equally, they coach the frontline staff through an immense change process. These managers are still essential, but their role has fundamentally changed.
Critical thinking skills may require some work. Consider the results of a 2023 NBER/National Bureau of Economic Research working paper by researchers from MIT, Harvard, and Purdue. AI working independently was found to be more effective at medical scan interpretation, diagnostic accuracy, and management reasoning than radiologists working with AI. Why is that? The study highlighted several biases the radiologists held against AI. They often undervalued the AI input compared to their own judgment, sticking to their guns even when the AI model proved to be correct. But the AI models had their own distinctive flaws as well. AI agents were far less effective than humans at gathering patient information in initial consultations, frequently failing to ask follow-up questions and missing contextual clues. While the report is damning about the effects of human biases, its major takeaway is not that human physicians should be replaced with “Robo Docs”; rather, it shows that AI is most effective when it is utilized in ways that take advantage of its own strengths and those of human doctors. Distrusting all of AI’s outputs is folly, but so is accepting them all as gospel.
Those who adapt to the new AI-infused model effectively will find themselves at the center of something powerful: organizations where intelligence doesn’t reside in a database or a boardroom, but is alive and evolving in the work itself.
A New Responsibility for Senior Leaders
Senior leaders, too, must reconsider how they lead in this new paradigm.
A century ago, ship captains had wide discretion to make decisions on their own. But with the invention of radio, that autonomy quickly evaporated. Now, admirals could peer into distant operations and issue direct orders, so they did. Technology enabled centralization, and leaders are fond of leading from the front.
AI may become the radio of the modern enterprise. With dashboards offering real-time visibility into everything from performance metrics to team sentiment, executives may be tempted to micromanage more than ever before.
They must resist that urge.
Visibility should provide reassurance, not license for interference. The job of senior leadership in an Octopus Organization is to define intent, shape culture, and build systems that make decentralized decision-making safe, smart, and aligned. It is not to dictate from the bridge. If a shipwreck looks possible, then they can intervene in a timely way.
How can senior leaders ensure that decentralization meets all its goals? Figure 1 shows four actions leaders should take in leading decentralization.
For distributed intelligence to function, the parts must be connected. The octopus’s arms coordinate through a bundle of nerves that scientists call a neural necklace. One arm can orchestrate with the other without needing to involve the central brain. Organizations need something similar.
Leaders need to design systems where insight travels easily across teams, silos, and geographies. This ensures that what one part of the organization learns, others can apply. Leaders must champion the infrastructure—technical, behavioral, and cultural—that allows knowledge to flow and grow.
In practice, this means removing friction: redundant approvals, information hoarding, outdated key performance indicators/KPIs, or clumsy workflows. It means designing for transparency, trust, and learning.
And it means investing in data not mainly as a reporting tool, but as a shared resource for judgment and innovation.
Of course, just because you can collect and distribute data doesn’t mean you should. Most employees don’t need to know their colleagues’ salaries, for example. Data transparency costs more than it delivers if the data isn’t used in the right way. Employees don’t want to feel as if their every keystroke is monitored. They may also balk at the additional administrative lift that collecting and sharing certain metrics imposes on them, such as delivering constant project status updates.
Employees don’t want to feel as if their every keystroke is monitored.
Much of the important information leaders require to make key decisions can’t be measured. Data collection and distribution involves judgment calls. It’s easy to mistake a proxy for the thing it’s being used to measure. For instance, tracking keystrokes and mouse movements is used to measure call center productivity. This KPI incentivizes employees to conduct busy work that produces more keystrokes, when they could be using the time to find better solutions for customers. Efficiency metrics often discourage employees from seeking advice from experienced managers, leaders, and contributors, whose stories may be unknown to AI.
AI systems are only as good as the data they process—if too much data is suppressed, they can’t do their jobs well. Whatever you do, prevent the mantra “Measure What Matters” from turning into “Only What Can be Measured Matters.” AI may confidently encourage you to make the wrong investments and draw the wrong conclusions. It can limit the range of signals your organizations relies on to track progress. Take one historical cautionary tale; during the Vietnam War, the Pentagon collected reams of data on how many enemy soldiers were killed, yet the clear qualitative trend was that the United States was losing the war.
To win in a world where so much is measurable, it’s important to be clear on what matters and prioritize that.
Avoiding the Cultural Cliff
Not all reactions to AI will be enthusiastic. Poorly implemented, AI can make employees feel diminished, reduced to the role of executor and following a machine’s lead. This undermines both morale and the very human judgment that organizations still need.
Leaders must communicate clearly that AI is here to augment, not replace. It is a career accelerator, not a career ender.
The Octopus model only works when people believe they are part of something intelligent and purposeful. They need clear principles to direct their everyday decisions, alongside the AI algorithms. Amazon has articulated 16 leadership principles that guide over a million-plus employees worldwide. An organization that vast would struggle to maintain coherence without principles that can ensure common ways of working. Principles become even more critical as the pace of business rapidly accelerates, which is one surefire thing about the AI Age.
Leading the Transformation
The transformation to an Octopus Organization will not happen overnight. It requires courage, experimentation, and sustained leadership. But the alternative—clinging to an outdated model while the world races ahead—is far riskier.
Leaders who move now can reshape their organizations to thrive in an AI-driven world. Those who delay may find themselves trapped in models that no longer serve.
Take a fresh look at the work that your organization needs to get done. What really needs to happen, and how can AI be leveraged to best help humans accomplish that? For instance, if one piece of work is to configure pricing to fit a customer’s sweet spot, how might AI assess the customer’s needs and their willingness to pay, balancing those against the underlying costs of serving that customer? How might humans ensure that the terms are appropriate and then sell the product to the customer? What might an AI-enabled process save in terms of time and labor versus your old ways of doing things? With this rethinking of the process, who would be doing what? What capabilities—AI and skill-based—would they require to succeed?
Microsoft is replacing its org charts with “work charts” focused on the work to be done, rather than on the seniority and supervisory authority of managers. You can do the same for each of your functions. Go through them systematically and specify all the jobs within them, defining them not in terms of what humans currently do, but as discrete chunks of what must happen to keep the organization running smoothly.
Could some of those jobs be automated? Taken over by external partners? The answers could dramatically expand your possibilities.
Overall, begin with:
- Identify key decisions that can be redistributed
- Define bands of acceptable risk for devolved decisions, and boundaries
- Invest in providing middle management with the skills they will need
- Build a “neural necklace” for organizational insight
- Shift executive focus from control to orchestration
Above all, lead with intention. The future doesn’t belong to those who simply add AI to the status quo. It belongs to those who lead differently—who grow arms, not armor.
