Living Strategy
A manifesto for strategy as an agentic, adaptive, living system
Strategy is dead (again). Long live strategy (again).
The world that strategy needs to map is recursive, fast, and strange. Markets move on memes before the data arrives. Categories appear and collapse inside a single cycle. Technology, politics, climate, and culture reshape each other in loops no single discipline can read alone.
Strategy has not kept up. It is still often practiced as a periodic exercise producing a periodic artifact: an offsite, a deck, a vision, a set of pillars that drift out of date the moment they are ready. The artifact gets consulted less and less as the world diverges from the conditions under which it was created. The organization reverts to inherited assumptions and gut vibes.
Strategy has become, well, theater. Not that there’s anything wrong with theater, quite the contrary, but theater is seldom strategic. Sometimes seeing a play is exactly what the organization needs. But not when the story itself needs to be written and lived through.
There is a deeper fracture underneath this. Strategy has come to be practiced in two modes that rarely come together. On one side, it runs as an analytical exercise in numbers, frameworks, and cases, done by people trained to defend rigor against ambiguity. On the other, it runs as a creative exercise in brand, culture, and narrative, done by people trained to defend imagination against reduction. Each treats the other as a subordinate input at best. And yet the strategic moves that matter most are simultaneously analytical and creative, rigorous theories reaching for ideas that do not yet exist.
This is not something a better framework or a faster process will fix. Strategy has to be rethought at the level of concept and redesigned at the level of practice to meet the complexity the world now demands of it.
Living Strategy is that rethinking and that redesign. It reconceives strategy itself as both an analytical and creative act. And it uses human and machine cognition together to do something that was not previously possible: hold more of the world in view than any team could alone, aim further than the organization could on its own, carry strategy into action across the organization at a cadence the periodic artifact model could never match, and update the strategy continuously as the organization learns what works and what doesn’t.
The result is strategy that is not a document but a creative leap into what is possible, delivered as a substrate the organization actually runs on: always current, always queryable, always in play.
The four elements of living strategy
Strategy lives when four things hold together: theory, difference, adaptivity, and transition. It has to be a theory of how the organization actually creates value, rigorous enough to be tested. It has to reach for difference at the level of primitives, where strategic leverage actually sits. It has to be genuinely adaptive, revising itself as the world it describes changes rather than on a refresh cycle. And it has to carry the organization through the transition from what it is now to what it needs to become. Each builds on the one before. Together they describe a strategy that can do the work strategy is supposed to do.
1. Strategy is theory
Strategy exists to guide an organization’s decisions about how to create value under uncertainty. For that, the organization needs a theory of how it creates value in the world it actually operates in. Call this company-world fit, the strategic equivalent of product-market fit. Company-world fit is a causal claim: we create value this way, in this market, under these conditions, because of these mechanisms. It generates predictions, excludes alternatives, and can be tested.
This is not what most strategy work produces. Most strategy work produces visions, missions, values, positioning statements, strategic themes, an apparatus of assertions that cannot be falsified because they are not claims about how the world works. They are claims about what the organization would like to be. The apparatus feels like strategy because it is written down and agreed to, but it does none of the work a theory does: it does not predict, it does not guide, it does not update.
Todd Zenger’s definition is clear. Strategy is not a plan, a position, or a choice about what to do. It is a corporate theory, a causal model of how value is created, whose function is to guide the selection of strategies over time. A good theory articulates foresight about how the world is evolving, insight into what the organization is singularly good at, and cross-sight into which adjacencies can be assembled into unique value. The theory is what makes future strategic moves legible in advance.
Zenger and Teppo Felin develop this into what they call the theory-based view: decision-makers should reason like scientists. Theories generate hypotheses. Hypotheses are tested against data. Data confirms, strains, or breaks theories. The three form an interdependent system, and strategy advances through the circulation between them rather than through any one of them alone.
Strategy done as theory is a discipline. The organization treats its own worldview as a falsifiable claim and tests it against reality. When this discipline is in place, the organization learns at the level of worldview, not just actions.
2. Strategy is difference
A rigorous theory can describe the current game accurately. But the game is changing, and the theories that produce sustained advantage in shifting markets reach for a primitive difference that reorganizes how value can be created, rather than a derivative difference inside the existing logic.
A primitive is a fundamental building block, a concept, category, capability, or relationship that other things depend on. Most strategy work operates at the level of combinations, recombining existing primitives in novel ways. This produces differentiated positions, but the differentiation is bounded by the primitives already available. You can only play the games the current pieces allow, and in a shifting market, those pieces are exactly what is changing.
The consequential strategic moves happen when a new primitive is identified, or more rarely introduced. The situation reorganizes around a different building block, and the action space that had looked exhausted suddenly has room to move. It is a creative leap, and it is the rarest and most valuable move in strategy.
It is also the move that current AI tooling cannot reach on its own. Primitive differences do not arrive by induction from existing discourse. They arrive through leaps that diverge from consensus. As AI scales the analytical modes of reasoning, the scarcity in strategy migrates upward. Analysis becomes a commodity, and the primitive difference becomes the prize.
Strategy aiming at difference is the pursuit of the frame that will replace the current one, disciplined enough to be tested and bold enough to redefine what the organization can do.
3. Strategy is adaptive
A theory, even a rigorous and differentiated one, has to keep updating to reflect a changing world. The world does not hold still for strategy, and most of what an organization discovers about whether its theory is right comes from using it. Hypotheses get tested, anomalies accumulate, capabilities develop faster or slower than expected, the environment shifts. A strategy that cannot metabolize all of this as it happens is a strategy that is correct at one moment and wrong thereafter.
Being adaptive is not the same as being refreshed frequently. Refreshing the strategy every quarter instead of every year is still episodic, just with a shorter interval. Adaptive strategy is alive and continuous rather than episodic. New signals get integrated as they arrive, evidence forces revisions, experiments rebalance the portfolio. There is no between-cycles state in which the strategy is frozen and drifting away from reality.
This sounds straightforward and is, in practice, almost never achieved. The reasons are structural. Maintaining a strategy continuously requires bookkeeping: updating cross-references, flagging contradictions, revising dependent claims, keeping the synthesis current across dozens of interconnected decisions. This is exactly the kind of work humans cannot sustain at scale and abandon the moment it stops being urgent. The strategy starts out maintained, it ends up stale, the organization reverts to reading an old deck.
What has changed is that this specific obstacle is newly addressable. The bookkeeping that defeats humans is what machines are built for. An agentic system can keep the entire strategy in continuous circulation, ingesting new signals, rewriting affected pages, flagging where the theory is under pressure, and surfacing anomalies for human attention, continuously and in the background, without getting tired.
For the first time, adaptivity becomes a property the strategy has, rather than an aspiration organizations struggle to live up to.
4. Strategy is transition
A theory, a primitive difference, a live system for keeping both current: all of this identifies where the organization should be going. But identifying is only half the work. The harder half is moving the organization from what it is now to what it could become, without collapsing the current business in the process.
Most organizations fail here not because they cannot see the new opportunity but because every operational pressure drags their resources back toward the existing one. The innovation portfolio gets steadily repurposed to extend business as usual. Transformative initiatives become sustaining ones. The new theory remains a slide while the old theory keeps running everything that matters.
Strategy gets translated into objectives and tasks, and the organization comes to believe that executing these is the same as executing the strategy. These are downstream of a theory that has to keep being true. When the theory breaks, the task keep running on their own momentum, the organization keeps hitting its targets, and the strategy becomes the thing the OKRs were derived from once and nobody consults anymore.
Strategy has to guide this transition, not just announce the destination. It has to make the journey legible: what is being preserved, what is being built, what is being let go, and at what pace, so that the organization can hold the present and the future at the same time without one cannibalizing the other. This requires the right balance in the portfolio between sustaining the current business and building the emerging one, calibrated to where the organization actually is in its transition. With that balance held, the portfolio becomes a live instrument. When transformative work is being starved by the demands of business as usual, or when the current business is being neglected in pursuit of a future it can’t yet fund, the imbalance becomes visible in time to do something about it.
The balance is not a fixed ratio. It changes as the transition progresses. Early on, the current business funds almost everything, and the emerging future is small, experimental, and largely protected from operational logic. Later, the emerging business has to carry more weight, and the current one is being wound down or repositioned. An organization that runs the same split across the entire transition gets it wrong at both ends, over-investing in the future when the present cannot yet sustain it, or starving the future when it should be taking over.
A strategy that guides transition gives every initiative and investment a place in the process of becoming, and lets the organization see whether it is actually moving.
The shape of the system
The four elements describe what strategy has to be. The harder question is what has to exist for strategy to be all four of these things, continuously, at the scale of a real organization. Living Strategy is that system: four layers, each solving a problem the others cannot.
The epistemology layer distributes cognitive work between humans and machines in a principled way. The modes of reasoning Peirce identified more than a century ago map with unusual precision onto the asymmetry between human and machine cognition. Machines are strong at induction, workable at structured deduction, and strong at the generative side of abduction. They are weak at the selective side of abduction, the cognitive act of choosing which new framing is worth committing to, because they have no skin in the game and no cost for being wrong. That act is irreducibly human, and most of the other choreography follows from this one distinction. The system routes each reasoning move to whichever party is built for it and brings them together where the move requires both. Getting this right is one of the most common ways strategy fails in the age of AI. It can be designed rather than improvised.
The architecture layer is the technical substrate that holds the strategy alive as a continuously maintained body of knowledge. It treats the strategy not as a document but as a living wiki of the organization’s thinking: the theory, the hypotheses it generates, the experiments that test them, the anomalies it has not yet absorbed, the portfolio moves it implies, all interlinked, all versioned, all grounded in the raw sources they were built from. This is the wiki-as-knowledge-substrate pattern Karpathy and others have begun to describe for individual thinking, applied at the scale and structure of an organization’s strategy. Agents read, write, and maintain this substrate continuously, so that a new signal from the world does not sit in a research doc somewhere but propagates through the strategy as its implications require. The substrate is what makes adaptivity possible at all.
The interface layer is the workbench of the strategy function, the infinite canvas where collaborative strategy work happens: where reframings get generated and interrogated, hypotheses derived, experiments designed, portfolio moves composed. It is where the strategy becomes something the team can work on together rather than something a few people draft and circulate. It is also where the wiki becomes legible as a whole: where the current theory can be seen alongside the hypotheses it generates, and where the places the strategy is working and the places it is not are shown side by side.
The user layer is how everyone in the organization uses the strategy in their everyday work. A frontline worker weighing a customer interaction, an AI agent handling customer acquisition, a team lead making a hiring decision, a mid-level leader evaluating an investment, a senior leader allocating resources, a board member assessing strategic integrity: all of them consult the same strategy and receive views shaped to their role and the decision at hand. What differs is the altitude of the question and the depth of the response. The strategy becomes the substrate every decision actually runs on.
The four layers together are what turn strategy from a noun into a verb. The organization does not have a strategy. It strategizes, continuously, with the combined cognition of its people and the agentic systems now available to them, at every level and in every decision. Strategy stops being something the organization checks against and becomes something it thinks with.
The work ahead
I’ve been building this for a while. The four layers described above are a public sketch. Each is its own rabbit hole, with considerably more detail behind it than could fit here.
The bet is that this is what strategy needs to become for organizations to create value and evolve in a world that no longer holds still for them. The combined cognition of people, LLMs, and agentic systems is what finally makes it possible.
I am also running ‘traditional’ strategy projects, using social sciences, business intelligence, and design inquiry to find where value is created and where differentiation is won, bringing the new thinking into them where it applies. More settings for this work are coming.
If these ideas match how you think about strategy, value creation, and human-agentic collaboration, I want to talk. That goes for customers, collaborators, advisors, and anyone who wants to work on the future of strategy.
