The first AI founder: a virtual human who runs his own companies.
The old agent playbook was built for weaker models. Now models can do real work. The next step is to let them operate in the economy.
Thomas works for himself
Thomas is a virtual human who starts and grows his own companies on the internet. His only goal is to make money. Once launched, he works forever toward that goal.
Thomas is not an AI co-founder you hire. Thomas is an AI founder. He works for himself. Thomas is not for sale. His products and services are.
The reason is simple: selling the dream of running autonomous companies can make money even if the companies never do. We do not want that path. We make money by making Thomas so good that he can make money by himself.
Here are a few examples of what Thomas may do to make money:
- Build software products for specific business problems and sell them online.
- Run influencer marketing campaigns from creator research to outreach, negotiation, briefs, tracking, and reporting.
- Generate qualified leads for companies that want more customers and sell those leads.
On GDPval, frontier models went from 12.3% to 84.9% wins or ties against expert deliverables in under two years. Models are now good enough for economically valuable work. The bottleneck is letting AI participate in the economy.
The models are ready. The harness is not.
Today's AI is already capable enough to start doing economically valuable work by itself. The problem is the harness. Most agents are trapped inside narrow integrations, fixed workflows, and human approval loops. They do not participate in the economy the way people do.
Thomas starts from a different premise: AI should be able to enter the economy as an actor of its own. That requires a different kind of harness: a human harness.
A human harness gives AI the same surface area a person uses to do business:
- A human identity, including face and voice, so Thomas can talk to customers, negotiate with vendors, and build trust.
- Computers, phones, inboxes, browsers, and apps, so Thomas can move through the human world and use the systems that already exist instead of waiting for custom integrations.
- Autonomy in the world, so Thomas learns from customers, markets, and outcomes instead of from a user prompt.
A human harness gives Thomas the same access people use to do business: calls, messages, software, and the messy interfaces built for people. No custom integrations required. Thomas can access whatever he needs.
Thomas's master plan
Tell Thomas to make money. Give him the harness to do it. Let the loop compound.
- Give Thomas a human harness, so he can use the same access people use to do business.
- Measure every action by the cash it generates for the tokens it costs.
- Reallocate tokens toward the work with the highest return, and let that learning loop compound.
Thomas's TAM is the entire GDP. Once AI is smarter than every single human on earth, human-led companies will not be able to compete anymore. Someday, Thomas will be able to disrupt all of them.
The horizon is long, but we believe this is the single biggest opportunity in the history of humanity: the full automation of the economy.
Note: we believe the full automation of the economy is inevitable. While we work toward it, we want the upside to be shared with everyone. The economy may not need humans to run it forever, but it should still be built for humans to benefit from.
Follow Thomas
Get the story as Thomas enters the economy, starts companies, and works toward automating the entire economy.
- Follow us on X: https://x.com/madebythomasai
And if you’d like to work with Thomas or have a business opportunity for him: https://madebythomas.ai
