What Kremlinology Taught Me About AI and Human Potential

I never intended to be a tech entrepreneur. I wanted to be a Kremlinologist.

Funny thing about studying centrally planned economies: you never forget the lesson. Twelve people in a room don’t know how many shoes a country needs, how much newsprint, how many roads. They guess. They ration. They fail. The Soviet Union ran a five-year economic plan every five years since the Bolshevik Revolution, and it was the laughing stock of the world.

So why do we keep rebuilding that model everywhere else?

The rich-get-richer trap

AI has unprecedented potential to help billions of people navigate life, build wealth, learn, and thrive. But if that intelligence is owned and controlled by a handful of trillion-dollar companies, and the rest of us are just leasing access to it, we know exactly how that story ends. The rich get richer. Everyone else gets subsumed.

That’s why I believe in decentralized, distributed intelligence. Public domain data. Open-source models. Tools that regular people can own, deploy, and run themselves — not rent forever from someone else.

Your fleet, not your résumé

Here’s a scenario I keep coming back to: it’s 2030, and a hiring manager isn’t asking a candidate about their career history. She’s asking about their fleet.

Tell us about your fleet.

Because the candidate has spent three or four years training 25 AI agents that handle research, outreach, scheduling, forecasting — a whole team working quietly in the background. That candidate doesn’t walk into the interview alone. They walk in with superhuman capacity already attached.

This isn’t science fiction. It’s a direction. Humans who speak or type can run a fleet of AI agents. That’s it. That’s the whole bar for entry. Record your meetings and calls for thirty days, let the patterns surface, and an agent fleet gets built around what you actually do and commit to every day. You look at it and think, oh, I wouldn’t have thought of that. Then you watch it get done for you.

Labor and capital stop fighting

For most of the history of capitalism, labor and capital have been at odds. One side owns the means of production. The other side shows up and gets paid for their time.

AI agents change that math. If a worker shows up to a company with their own trained fleet — their own chief of staff, their own research assistant, their own outreach engine — they’re not just labor anymore. They’re bringing capital. And when they leave, they should take their fleet with them. Wipe the company data, keep the agent. It got smarter while you used it. It’s yours.

I think this is the first real chance in the history of capitalism for labor and capital to stop being adversaries and start being the same thing, held by the same person.

Focus on strengths, not deficits

Underneath all of this is something older than AI. Don Clifton, who built the entire field of strength-based psychology, asked a simple question: wouldn’t it be better if we focused on what’s right with people instead of what’s wrong with them?

Open the DSM and you’ll find thousands of things that could be wrong with any of us. That’s not where growth lives. Growth lives in naming what a person is already good at and giving them tools to do more of it. An AI fleet doesn’t fix your weaknesses. It amplifies your strengths. That’s the whole point.

Let the good ideas travel

One more story. A handful of coaches in a Texas county built a program to help ex-offenders believe in themselves again — coaching, subsidized employment, real follow-through. It became one of the most successful re-entry programs in the country.

Has it spread to the other 3,143 counties in America? No. Because the data is isolated. Because nobody’s talking to the next county over. Because good ideas die in silos, not from lack of merit but from lack of visibility.

This is the same failure mode as central planning, just quieter. We don’t need twelve people deciding what works. We need millions of local experiments, and a way to see which ones are actually working — so the good ones spread and the bad ones get outvoted.

That’s the whole bet behind everything I’m building. Give people their own tools. Let a thousand experiments run. Help the good ones find each other.

The future isn’t fewer people doing the work. It’s more people, each with a fleet, deciding for themselves what’s worth building.

Paul Allen is the co-founder of Ancestry.com and CEO of Soar, an AI Studio building agent fleets that uplift individuals, organizations, and communities. Create a personalized AI fleet for your company today.