At AI Innovation Week in Mexico, we explored a counterintuitive idea: As AI makes public knowledge cheap, relationships and unspoken knowledge rise in value.

In 2001, as the energy-trading firm Enron collapsed in scandal, most investors ran the other way.

Not Ken Griffin.

The Citadel hedge fund founder chartered a Gulfstream and sent sixteen researchers to Houston with a mission. He sent them to interview 300 former Enron employees. He wasn’t looking for dirt. He wanted to uncover how Enron’s legendary energy trading desk actually operated.

These conversations unearthed something you couldn’t find in a manual or a website: the informal systems, personal judgments, and off-book patterns that powered one of the most advanced trading operations in the world.

With that invisible playbook, Griffin built a new energy trading arm. It made him hundreds of millions.

What Griffin understood—and what matters even more as AI permeates our lives—is that the most valuable expertise isn’t found in manuals or databases.

It lives in three distinct layers.

Not All Knowledge Is Created Equal

Griffin’s success points to a crucial framework for the AI era: not all knowledge is created equal.

Domain Knowledge

Domain knowledge is the published facts, formulas, and regulations anyone can look up. This is rapidly becoming AI’s territory. If it’s been published, digitized, or taught in school, AI can probably access it better and faster than humans.

This type of knowledge will get hoovered up into LLM training data with ease and increasing frequency. The two below, however, will not.

Institutional Knowledge

This is the unwritten organization-specific know-how: which risk limits actually matter, whose approval really counts, and which metrics truly drive decisions. This is the “how things really work around here” layer that gives established players their home-field advantage.

Tacit Knowledge

Tacit knowledge comprises the intuitions and muscle memory born from experience: a trader’s feel for when market sentiment shifts, a chef’s adjustment to cooking temperature based on smell, a negotiator’s ability to read body language. These skills are nearly impossible to articulate, let alone code into an algorithm.

As AI democratizes access to domain knowledge, competitive advantage is shifting decisively to the inner layers. Companies and people that systematically identify and protect their institutional and tacit knowledge are the ones that will thrive as AI creeps into every facet of our work lives.

But Griffin’s real advantage wasn’t just knowing what knowledge to capture. It was knowing who to talk to.

Why Networks Still Matter

Griffin’s Enron expedition reveals something technology can’t disrupt: the power of relationships. The entire knowledge transfer operation, worth millions, hinged on Griffin’s ability to make the right calls, open the right doors, and establish the trust needed for candid conversations with former Enron employees. Without these relationships, he never would have had the opportunity to interview them in the first place.

While we can’t predict exactly the exact way AI will change everything, I’m willing to bet heavily on human connections not losing their value. Cultivating relationships, both internally and across their industries, gives you a hedge against an uncertain future.

In the age of infinite information, the rarest skill is knowing who to call.

Thanks to early readers: Matthew Beebe, Cansafis Foote, Miche Priest