Allocating intelligence
Attention as the new discipline for the age of cheap cognition
In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients. Hence a wealth of information creates a poverty of attention. — Herbert A. Simon, Designing Organizations for an Information-Rich World (1971)1
The best investors and the best executives share one quiet habit: they decide where not to spend. That was the lesson of The Outsiders2: eight CEOs who outperformed, not by chasing growth, but by allocating with precision. They knew that scale without discipline destroys value, and that a lot of what looks like progress is simply motion.
Today’s AI brings a new focus to the same realisation. Intelligence is now abundant and almost frictionless to access. The temptation is to deploy it everywhere. Companies rush to “scale AI”, treating cognition like a commodity, flooding processes with agents and dashboards. But the real leverage doesn’t come from scaling intelligence; it comes from allocating it.
In Graham and Buffett’s terms, the intelligent investor3 protects downside first. The intelligent allocator does the same. Treat compute, time and attention as scarce capital. Leave room for error by investing where there’s a margin of clarity — use cases with a well-defined problem, known workflow, and measurable return. The goal is not to automate everything, but to de-risk the organisation’s thinking.
The most effective AI projects often look more like value investing than venture bets. They compound quietly. A retrieval tool that saves a team hours of searching, or compresses a two-week task into a day, often outperforms the glamorous moonshot experiment. DEC’s expert system showed this long before LLMs: 90% accuracy can be worth tens of millions when applied to the right bottleneck.
The Outsiders also understood that assumptions drive outcomes. The same rule now applies to intelligence itself — every prompt, workflow, and agent is a capital decision. Buffett’s advice still holds: be fearful when others are greedy, and greedy when others are fearful. Translated to cognition: be cautious when others automate blindly; be curious when others pause.
In this climate of change, returns are likely to follow a power-law distribution. Most experiments will fail. A few will redefine whole categories. The companies that prosper will share one trait: optimistic focus. They’ll hold two assumptions at once:
Risk must be minimised
The future will surprise us
The discipline isn’t to eliminate uncertainty; it’s to stay solvent and curious long enough to benefit from it.
Conventional view: What jobs can we give the intelligence?
Outside view: Where will the next unit of intelligence yield the highest return?
Intelligence is cheap. Clarity is the real scarcity.
Practising allocation
Start from the assumption that most AI initiatives will fail, and design a system that limits the cost of those failures while keeping exposure to asymmetric upside. That’s what the Outsiders did: they protected cash, made small reversible bets, and doubled down only when the evidence was unambiguous. The intelligent allocator behaves the same way.
Three grounded principles for allocating compute, time and attention:
1) Protect downside
Avoid large, irreversible commitments like multi-year programmes, untested integrations, and vendor lock-ins. Be prepared to abandon sunk costs, too. Good Strategy / Bad Strategy4 suggests that adaptability is better than dogmatically sticking to decisions — the most effective allocators aggressively cull projects that consume scarce technical and political capital.
If a workflow or vendor is structurally leaky, it’s usually better to switch boats rather than patch endlessly. Value compounds only when clarity does. Too much patience wastes time. Too much optimism wastes capital.
Instead, pursue a margin of clarity by launching experiments small enough to fail cheaply, but precise enough to reveal signal. A week spent validating a real workflow beats a year of building a platform no one needs.
2) Identify asymmetries
A few well-aimed use cases will return most of the value. Think of an internal knowledge tool, a pricing copilot, or an assistant that generates structured reports. The intelligent allocator doesn’t spread their bets evenly; they search for steep parts of the value curve where a small injection of intelligence displaces large amounts of wasted effort.
One must also be willing to defund yesterday’s success, as Outsider CEOs did, redeploying cash from profitable divisions. Productive allocators redirect intelligence from routine efficiency towards fresh understanding.
3) Compound clarity
When an experiment works, the real asset isn’t the model — it’s the understanding of why it worked. Each success sharpens the organisation’s ability to frame problems, set boundaries and integrate humans in the loop.
Clarity makes every form of capital compound faster — multiplying returns across future decisions.
The irony of the AI era is that the most rational way to scale intelligence is not to chase scale at all. It’s to treat every unit of cognition — human or machine — as an investment, and to seek opportunities to spend where it produces compounding clarity.
The companies that win won’t be the ones that deploy the most AI; they’ll be the ones who allocate it best.
The discipline of temperament
Isaac Newton could calculate the motions of the planets, but not the madness of people. The discipline of allocation is more about temperament than intellect — the self-control to hold fire when the crowd rushes, and to explore when others freeze.
The Outsiders treated each move as if testing a hypothesis. Rumelt’s kernel of strategy might describe this experimentation process as diagnosis → guiding policy → coherent action.
Today, AI lets us implement strategy fast. Each prompt, workflow or agent is an empirical test of how we suspect the world works. Leadership’s job isn’t to publish a plan and hope for the best. Instead, it’s to design the portfolio of experiments that can prove it right or wrong quickly.
The best allocators are critical and opportunistic. They’re cautious until conviction strikes, and decisive when it does. They think like foxes, not hedgehogs: holding multiple models lightly rather than betting on one big idea. They spend frugally, dwell in System 2, and resist the market's mood. Like Adam Smith described in The Wealth of Nations5, frugality is the quiet foundation of all capital formation.
Intellect isn’t enough; temperament separates high-signal compounding from the waste that comes from the most common distractions.
The rise of the intelligence allocator
A new archetype defines how value is created in each era:
Industrial age: operators → apply machinery
Digital age: optimisers → apply data and scale
Intelligence age: allocators → decide where cognition should flow?
Intelligence allocators won’t necessarily need to write code or train models. But they will design how and where intelligence is deployed: not choosing between human and machine, but defining how the two reinforce one another. They’ll treat compute, time, and attention as capital — resources that compound when aligned with clarity.
The CEO’s job hasn’t vanished. The same framing of operate, fund and allocate still applies to the substrate of intelligence. Operations run on code, funding increasingly goes to compute, and allocation decides where cognition is spent. The right leadership move isn’t telling AI-native knowledge workers how to prompt; it’s creating the conditions for intelligent systems and people to allocate themselves effectively.
Every company needs at least one person who can tell the difference between an experiment and a strategy. Someone who knows “AI at scale” isn’t the goal, but a by-product of clarity. Titles may change. Tools will definitely change. But the winning discipline stays the same: allocation over execution.
The next generation will master the allocation of intelligence and the stewardship of attention.
Designing Organizations for an Information-Rich World in Martin Greenberger (ed.) Computers, Communications, and the Public Interest — Herbert A Simon (1971)
The Outsiders : Eight Unconventional CEOs and Their Radically Rational Blueprint for Success — William N. Thorndike Jr. (2012)
The Intelligent Investor — Benjamin Graham (1949)
Good Strategy Bad Strategy: The Difference and Why it Matters — Richard Rumelt (2011)
The Wealth of Nations — Adam Smith (1776)