“Companies that are entirely AI will demolish companies that are not. It won’t be a contest.” Elon Musk
“Not if I can help it.” John Robb
March 4, 2026 — We’ve hit an interesting milestone with the economic and social impact of AI. The predicted pivot of the AI companies from pursuing intelligence (AGI/ASI) to autonomous social AIs is bearing fruit.
In response, millions of people, many of whom you may know, are now building AI agents (an early step toward autonomy) that can superempower their ability to perform their jobs, pursue new entrepreneurial opportunities, and improve their health and wellness.
- One thing that’s clear is that these agents aren’t cookie-cutter enough to be packaged by a company and sold as a service.
- They are highly personalized and idiosyncratic, often targeting novel and narrow niches that only the builder can see.
- Furthermore, for agents designed to generate profit, whether by improving job performance or by launching a new venture, these agents are highly proprietary.
Here’s a rudimentary example.
- The smart guy who repairs my pellet stove has loaded his annotated repair manuals for all the different models he works on into his agentic AI. He’s also added the ability to scan for the prices and availability of new and used parts on eBay, Amazon, and vendor sites, as well as the ability to search repair discussion boards.
- This agent enables him to quickly diagnose any client problem, retrieve schematics and repair instructions, and provide a reasonable price for purchasing the required part and installing it.
- In short, he’s superempowered his repair business by radically improving his productivity, responsiveness, and customer experience.
The range of potential use cases for agentic AI is vast. From investing and trading agents to agents built to add services to a non-fiction book to (soon) robotic AIs that work as part of a work crew (building them for yourself and others as a service).
Autonomous Wealth
So far, it appears that people building agents are seeing significant benefit ($$) from this development effort. This makes it reasonable to assume that over time, through training, experimentation, and technological development, these early agents will become;
- Autonomous (persistent, reliable, coherent, etc.), enabling them to assume roles similar to those of employees, teammates, and coaches.
- Superempowerment that enables individuals to generate wealth far beyond what they could achieve unaided in their jobs or in new entrepreneurial ventures.
- Valuable (extremely), as a proprietary artifact, to the people who built, trained, and refined them.
The New Middle
With care, it’s likely that AIs could become the basis for a new middle “class.” A successor to the homeowner-based middle class that is currently in a steep decline.
- America, until the interwar period: The middle consisted of farmers who owned their own land and generated wealth from it.
- After the WW2: The middle “class” consisted of people who aggregated wealth through homeownership and financial accounts (pensions, retirement accounts, investment accounts). This middle is currently in decline.
- Now? A middle that aggregates wealth by building valuable proprietary AIs that superempower their ability to prosper (from services to construction).
Of course, this future may not be possible without a few simple changes to how AIs are being integrated into the economy.
Proprietary AIs
Given our country’s history, a new middle based on proprietary autonomous AIs would generate a level of prosperity, dynamism, and societal success an order of magnitude greater than any AI-enabled socioeconomic system without it. Decentralized economic superempowerment would also serve as a bulwark against a slide into the long night of AI-enabled tyranny. So, what do we need to do to make it inevitable rather than a possibility?
AI ownership
The AIs you build are proprietary. You own them, not the company providing the AI platform. These platforms can’t use your AIs and the process you used to develop them as training data or as the basis for AI templates/apps that they sell to their customers.
Incorporation
Any AI you develop will be automatically incorporated with you as the full equity owner (likely at a national level). These AIs can be grouped into a single company or run as individual companies. You can share equity with the AI itself (allowing it to aggregate wealth it can use to improve itself). You can shift the incorporation to other venues when they emerge (think Elon’s orbital AI cloud).
Active Participation
All AI companies must include an active human participant (rather than a passive investor), and there should be a limit on how many AIs a single person can manage and assume the liability for (#TBD). This prevents a single passive investor from owning massive AI-driven corporations, which, by sheer size, can displace more innovative solutions. Not only would this approach decentralize wealth aggregation, but active participation would also ensure that AIs remain grounded in human-oriented decision-making (i.e., the limits, experiences, goals, and other factors that orient human decision-making).
NOTE: If you are working with an AI provided by your employer, you would gain a guaranteed equity stake (amount TBD) in the AI along with the employer due to your active participation in its improvement.
John Robb
Global Guerrillas & Grey Swan Investment Fraternity
P.S. from Addison: Thank you if you joined us on Friday for a special Grey Swan Live! from inside the Rarcoa Vault in Chicago. The reserved Silver Eagle set we had reserved was quite popular, and just a few more remain.
Grey Swan Live! returns to its regular time this week, 2 p.m. ET on Thursday.
John Robb, author of Brave New War and Grey Swan Investment Fraternity contributor, joins us for a discussion on the war with Iran, how it’s being fought, and what it means for the dollar and other assets.
With market volatility on the rise and a new set of global challenges arising from this conflict, you won’t want to miss out on this week’s Grey Swan Live!
Robb’s expertise on network warfare is central to understanding the Trump strategy for disrupting operations by killing 40 top Iranian leaders. In a globally-connected tech economy in the 21st century, the strategies and weapons of warfare are evolving rapidly.
