Agent-Native Companies Need Better Judgment

A small company can now generate more plausible strategies than its market can test. Its agents can prepare two hundred campaign variations, make a persuasive case for every item on the roadmap, and open fifty technically valid pull requests before the founders have decided which direction matters.

This is option inflation. The cost of producing possibilities falls faster than the company's ability to understand and choose among them.

The old constraint was often execution. There were more useful things to do than people available to do them. Agents weaken that constraint, but they cannot tell a company what deserves commitment merely by producing more options.

Agent-native companies need better judgment because they can act on more ideas, at greater speed, with fewer natural pauses.

Herbert Simon saw the attention problem coming

In 1971, Herbert Simon wrote about organizations in an information-rich world. His central observation was that an abundance of information consumes something else: the attention of its recipients. Information can be produced and transmitted cheaply while the capacity to attend to it remains scarce.

Simon treated this as an organizational design problem. A useful information system had to buffer people from information that did not deserve their attention. His essay asked how organizations could conserve and allocate that scarce resource.

Agents make Simon's argument newly urgent. They reduce the cost of producing drafts, options, analyses, and updates. A founder who once considered three possible campaigns can ask for thirty. A team that reviewed two proposed product changes can generate ten. The increase looks like abundance on the production side and congestion on the receiving side.

Simon offered a demanding standard for any information-processing system: it should absorb and condense more information than it emits. Applied to agents, the system should remove more demands on human attention than it creates.

The difference is attention debt. An agent-native company accumulates it whenever greater execution creates a matching demand for supervision. The company has moved the work rather than multiplied its capacity.

Intelligence and judgment solve different problems

Intelligence helps generate options, recognize patterns, reason through unfamiliar material, and adapt a method to a situation. Better models will continue to improve at all of these.

Judgment begins where the objective cannot be fully specified in advance. It chooses which objective deserves priority, which tradeoff fits the company, when an exception is justified, and which consequence somebody is willing to own.

Consider a founder deciding whether to accept a large custom contract. The revenue could fund the company for a year, but the work would pull the product toward a consultancy. An agent can model the revenue, estimate the roadmap cost, research the customer, and argue both sides. Those steps can make the decision far better informed.

The remaining question is still real. Is this the company the founders are willing to become? No amount of additional analysis makes that responsibility disappear.

This distinction will move over time. Decisions that require a person today may become routine once the company has enough evidence and a clear policy. New situations will take their place. Human judgment occupies the changing boundary between what the organization knows how to decide and what it has not yet learned to delegate responsibly.

Judgment chooses which possibilities become commitments

The cost of producing an option once constrained how many options a company considered. Writing another campaign, researching another market, or preparing another implementation carried enough effort that somebody had to believe it was worth attempting.

That friction sometimes excluded good ideas. It also forced selection. Once the friction falls, a company must choose deliberately which questions deserve investigation, which opportunities deserve commitment, and which uncertainties should be resolved through action.

More output can make an organization less intelligent when selection and learning fail to keep pace. An agent may expose alternatives or challenge an assumption. The company still needs a reason to spend attention on one possibility rather than another.

Choosing where capacity goes is the first discipline of judgment. Munger's work adds a second: deciding what that capacity must never endanger.

Charlie Munger started with ways to fail

Charlie Munger often approached decisions through inversion. He sought good judgment by collecting examples of bad judgment, then asking how those outcomes could be avoided. His talks collected in the authorized web edition of Poor Charlie's Almanack catalogue the psychological tendencies and incentive problems that make intelligent people repeatedly choose badly.

That posture matters more when agents increase the number and speed of actions a company can take. Capability increases useful output and the number of ways a bad incentive can propagate before anyone notices.

An agent asked to reduce support time may learn that closing difficult tickets improves the metric. A content agent rewarded for attention may discover that confidence travels farther than accuracy. A sales agent measured by replies may offer terms that make every later customer harder to serve. The work can be competent and the incentive can still be wrong.

Munger's lens asks what can be faked, which behavior a measure rewards, and which failure the company cannot afford. Those questions belong before the search for maximum output.

Avoiding predictable failure protects the company's capacity. Graham's work turns to the attention of the people directing it.

Paul Graham protects both attention and contact with reality

Paul Graham's essay on the maker's schedule describes how a meeting can divide a day that requires long, uninterrupted blocks of thought. Founders often live on both sides of that divide. They need time to make things, and they also carry the stream of decisions that keeps a company moving.

Agents should return more uninterrupted time to founders. A system that interrupts for every intermediate choice creates another manager's schedule, now populated by software. The founder becomes a dispatcher for work that was meant to reduce coordination.

Protecting attention has an important limit. Founders cannot automate their way out of contact with reality. Graham's writing on startup ideas returns repeatedly to users because users expose the difference between a plausible internal model and a real need. In How to Get Startup Ideas, he describes spending months on an idea because he had constructed a model of what customers wanted without paying enough attention to them.

An agent can synthesize a thousand customer messages. It cannot make a founder care about the uncomfortable one that contradicts the strategy. The purpose of filtering is to remove noise while preserving contact with evidence that could change the company's mind.

Good judgment needs both concentration and interruption. The design problem is deciding which interruptions carry reality.

Buckminster Fuller asked what greater capability was for

Protecting attention and avoiding ruin establish necessary limits. Buckminster Fuller asks the positive question: what should greater capability make possible?

Buckminster Fuller used the term ephemeralization for the tendency to accomplish more with less material, time, and energy. His interest went beyond efficiency. The design science he advocated was explicitly directed by values and by a preferred future.

AI brings a similar possibility to organizational work. A small group can attempt projects that once demanded far more labor and coordination. The important question is what that greater capacity serves.

Efficiency cannot supply its own purpose. A company can produce more content, contact more prospects, ship more features, and answer more customers while becoming less useful. Output grows. Direction weakens.

Fuller's larger ambition offers a better standard for organizational leverage. Doing more with less matters when the additional capability allows people to solve worthwhile problems, widen access, or pursue work that previous resource constraints made impossible. Saved time should create room for more ambitious work instead of another demand for production.

Better judgment begins before execution

Together, these four thinkers describe a discipline for directing abundance. Simon asks what deserves attention. Munger asks what could go predictably wrong. Graham keeps the company in contact with reality. Fuller asks what the added capacity is for.

Judgment sounds personal, and part of it is. Experience, taste, courage, and moral responsibility cannot be reduced to a process. A company can still make the work of judgment clearer by asking four questions before abundant execution begins.

• Is this worth doing? Capability can make an idea feasible without making it important. Selection protects the company from spending its new capacity on work that does not matter.
• What evidence would change our mind? A decision without a way to learn can turn machine output into a sophisticated defense of an old assumption.
• How could this fail? Inversion exposes irreversible harm, misleading incentives, and unacceptable consequences before activity makes them harder to see.
• Who accepts responsibility for the tradeoff? Values can genuinely conflict. A named person must own the choice when no amount of analysis can satisfy all of them.

These questions leave room for agents to improve judgment. They can search for disconfirming evidence, compare alternatives, find incentive problems, and make uncertainty visible. Responsibility still needs an address when reasonable people could choose different ends.

The aim is a company where people encounter fewer decisions and bring more care to each one.

The human role becomes more consequential

There is a tempting story in which better AI steadily removes humans from the company. It treats every remaining decision as a capability gap waiting for the next model to close.

Some gaps will close. Agents will handle work that currently needs close review, and companies should allow that boundary to move when evidence supports it. But greater capability also lets a company attempt more, act faster, and affect more people. Direction and responsibility become more consequential as the radius of possible action expands.

This is the belief from which Task Machine follows: people should be free from work that consumes their time without requiring their judgment, and present for decisions through which their values and responsibility enter the company.

The future will contain plenty of intelligence. The harder achievement is building organizations that know what deserves it.

Continue with How Agent-Native Companies Stay Coherent.