Every company will have to compound two kinds of capital — the people it grows, and the intelligence it owns.
01 — A different kind of platform shift
In every past platform shift, we used digital systems to enhance human capital — a one-way street. This transition is different. We can now close the loop: people and digital systems learning from one another, continuously. That changes how we even conceptualize work inside an enterprise.
What's at stake is not some tool and its use, but how organizations continue to learn, build IP, differentiate, and thrive in a world where AI models can continuously absorb the expertise of humans and organizations — and commoditize it.
“Without human direction, you have compute running in circles.”
02 — Two kinds of capital
The knowledge, judgment, relationships, ingenuity, and pattern recognition of your people. It sets ambitious goals, connects dots across domains, builds relationships, and recognizes the patterns that matter most.
The AI capability the firm builds and owns — its agentic systems, private evaluations, reinforcement environments, and queryable institutional memory. The compute that turns human direction into scale.
The counter-intuitive part: human capital does not become less valuable as token capital grows. It becomes more valuable. Human agency is the driver of token capital growth — the two compound only when they're coupled.
“You can offload a task, or even a job — but you can never offload your learning.”
03 — The learning loop
The firm's edge is a learning loop where human capital and token capital compound with every use. The author calls it a hill-climbing machine — and unlike most assets, it compounds. Every improved workflow generates better training signal, which accelerates the accumulation of tacit knowledge unique to the firm.
04 — Architecture for sovereignty
This needs a new architecture: every business builds agentic systems that improve over time while retaining control over its IP. The test of your sovereignty in the era ahead — can you switch out a generalist model without losing the company-veteran expertise built into your learning system?
Concretely, companies turn their workflows, domain knowledge, and accumulated judgment into systems that improve with each use:
Capture whether a model is actually improving against outcomes that matter to the business — not just external benchmarks.
Let models grow stronger on real traces from inside the organization, not generic public data.
Make institutional memory queryable, and make every token you spend more efficient.
This loop becomes the new IP of the firm.
“Build a frontier ecosystem, not just a frontier model.”
05 — What's at stake
The last thing any of us should want is a world where every company in every sector cedes value to a few models that absorb everything they see. There is no societal permission for an AI future that hollows out entire industries.
In the first phase of globalization, entire industrial economies were hollowed out by outsourcing. The GDP numbers looked fine on the surface, but the displacement was real — and the consequences are still being felt. Let's not bring that dynamic into the AI era, with a small number of systems capturing all the returns while industries find their knowledge commoditized right out from underneath them.
06 — The stable equilibrium
When every organization can own the learning loop that encodes its institutional knowledge, companies create value for themselves and the economy around them. Employees see their expertise amplified — their judgment becomes part of systems that make it replicable and scalable — and the benefits accrue to the companies and communities around them.
That is how firms drive value for themselves and the broader economy. And it is the stable equilibrium we should build together.