How Will People Make Money in the AI Age?

The host outlines 13 business opportunities in the agent era but rejects the premise that everyone must start a company to avoid being left behind; owning assets could also put people in a stronger position. AI-native service firms can use agents for most delivery and reserve human staff for exceptions. He estimates agents could handle 90% of the work: a bookkeeping firm, for instance, could have agents prepare records, charge per job or monthly, and rely on a small team to review unusual cases, potentially serving relatively few clients at high margins.

Distribution may remain difficult even as AI makes products easier to build, so the host sees value in owning an organic audience and community before developing products people ask for. He estimates Facebook ad clicks cost about $0.05 in 2009 versus $1–$5 or more today, depending on audience and country, while acknowledging he lacks exact current data. Advertising can help fund audience-building, and owners of multiple meme pages, YouTube channels, or Instagram audiences may benefit. Proprietary datasets are another opportunity when personal agents or ChatGPT cannot readily obtain the information; companies can acquire data and sell it to major AI labs. The host describes the data-acquisition boom as still in its first inning and cites Mercor and Micro One as data companies with billion-dollar valuations and hundreds of millions of dollars in annual recurring revenue.

The host distinguishes a model, which supplies the engine, from a harness: the industry-specific tools, workflow rules, memory, and checks that make a general model useful in practice. He sees room for systems such as an operating system for a freight-brokerage agent. He says about 90% of GPT-wrapper startups fail, but many of the top 10% evolved into harnesses as their models improved; Harvey and Spellbook are legal-sector examples. Vertical agents for recurring work also remain an opportunity. For independent restaurants, an agent could check stock, compare supplier offers, seek approval, and place orders.

Offline experiences can create businesses around in-person activities people want to share, though real estate can make them expensive to launch. The host cites Othership’s communal sauna and membership model, and suggests testing a lower-cost version first, such as a food truck or backyard launch. AI may also make niche hardware viable for underserved markets. Specialty coffee equipment shows that enthusiasts will pay for well-designed products, while The Minimal Company makes phones for people who do not want to scroll and reportedly raised millions of dollars on Kickstarter. The host expects niche products with dedicated followings and viral potential to find demand, alongside physical assets such as small industrial workshops and warehouses serving local tradespeople.

Robotics and physical AI opportunities may be nearer than many expect, and the host argues that selling hardware may be an effective way to sell software. Osmo provides window cleaning as a service by deploying and operating equipment for building owners, and reportedly raised about $6–7 million recently. Opportunities could include owning or renting equipment, as well as selling and marketing it, rather than building robots alone. Agent deployment also depends on scarce chips, electricity, cooling, and suitable compute locations, creating openings in infrastructure and energy; Crucio reportedly raised $3.9 billion for its vertically integrated AI infrastructure platform.

The host sees older adults as an underserved market for products and services that support happier, healthier lives, including care services, while cautioning that service businesses can be difficult to operate. He also sees room for a leading longevity business alongside Brian Johnson’s “don’t die” movement. Marketplaces and social networks can have strong moats, but reaching liquidity by matching supply and demand is hard; renewed momentum may come as personal agents seek access to platforms whose data is valuable. Mark Lew created TrustMR.com, a marketplace for buying and selling startups. Security is another opportunity: as agents handle sensitive work, businesses will need controls over what each system can access. An accounting agent, for example, could read invoices and draft entries without being allowed to access other data. The host expects more breaches at agent startups and greater demand for security, including services such as on-demand bodyguards with bulletproof cars.