Man's Search for Information

The enterprise has spent three years deploying AI and feels vaguely disappointed. The models work, the demos are compelling, the pilots impress — and yet, the magic that everyone experiences at home does not materialise at work.

This is not a coincidence. It is a design consequence — and understanding why requires looking at the history of how organizations were built to resist exactly this kind of change.

To see that clearly, we need to look at the longer arc: how each era of information access displaced the dominant constraint of the one before it.

Every technological era abstracts away the friction of the previous one. But the organizations built for that world persist — and often resist the transition long after the technology has arrived. The AI/Agentic Era is the same transition — except the resistance is no longer about access to information. It is about the identity of the people whose value came from discovering information and making sense of it.

The Great Contradiction

In our personal lives, the evolution of access is effectively complete. From the Library, to the Web, to mobile — each era collapsed the friction of the previous one. Now we don't search anymore. We ask, and intelligence arrives: synthesized, contextualized, and ready to use.

We have shifted from Pull (hunting for answers) to Push (intelligence arriving when needed).

The transition is already complete in your personal life. But the moment we step into the enterprise, the magic goes away.

The contradiction isn't that AI works at home but disappoints at work. It is that we have introduced technologies built for information abundance into organizations still designed around information scarcity. The technology has crossed the transition, while most organizations have not.

1. The Identity Crisis

Technology is rarely the hardest part. The harder part is what it changes inside the organization.

Model quality and hallucination are visible problems. The deeper problem is what AI does to roles, expertise, and authority inside the organization.

If synthesis and reasoning are automated:

• Who owns judgment?
• Who owns the outcome?
• What happens to the role that used to own the gathering?

No one volunteers to automate their role and depreciate their own utility. No existing title is incentivized to dismantle the structure that gives it power.

This is why AI transformation cannot be incremental. It requires redesigning the assumptions the organization was built upon:

• Redefining roles around outcomes, not tasks
• Re-architecting workflows so humans and AI trade control based on context
• Accepting that some roles, responsibilities, and identities will change

2. The Architectural Flaw

Enterprises today are attempting to bolt Push technology onto Pull workflows: systems that can surface, reason over, and act on information are being inserted into workflows that still depend on humans to pull that information together.

For decades, enterprise systems were architected around constraints:

• Data was scarce → We built silos to manage it (BI, Data Teams)
• Resources were expensive → We built processes to ration them (Tickets, Roadmaps)
• Context was fragmented → We hired humans to stitch it together (PMs, Middle Management)

That world is eroding. GenAI can reason across federated systems — email, CRM, contracts, logs — in a single pass. The constraint of missing context has theoretically collapsed.

But the architecture remains.

We still ask humans to provide information. We still design roles around retrieval. We still measure productivity through activity rather than outcomes. When you deploy AI into these workflows, you don't really transform work. You add a new layer of capability on top of assumptions that were never designed for it.

3. From Gathering to Judgment

In the Search Era, value came from gathering: finding the data, building the spreadsheet, writing the summary. Judgment was embedded in the process — who to trust, how to validate, which sources to give more weight to, what to leave out.

In the AI Era, information gathering is cheap. Machines can search, synthesize, and retrieve almost instantly. The scarce capability is no longer retrieval. It is judgment — the ability to interpret what arrives, decide what matters, and act coherently under uncertainty.

Personally, we already inhabit this shift. We probe, challenge, and refine AI outputs instinctively. Judgment is the work.

At work, most organizations still assume that finding information is the job. Roles, workflows, and measures of productivity remain organized around retrieval — even as retrieval has become the least expensive part of knowledge work.

Organizations built around retrieving information are increasingly optimizing the least valuable part of modern knowledge work.

The Real Transition

We often describe this moment as the transition from Search to AI. That framing is too small.

The deeper transition is from an organization designed around the cost of retrieval to one designed around the value of judgment. Those are not the same organization. One optimizes retrieval while the other cultivates judgment. Deploying better technology does not automatically create the second.

That redesign — not the technology itself — is the transition Doctrine exists to understand.

The real structural shift ahead involves three domains that remain largely unreformed: