AI Solo Businesses and Jobs: Count the People Outside Payroll

On November 2, IBTimes reported Claire Vo's account of running ChatPRD as its only full-time human employee, supported by nine AI agents. [1]

I take that seriously. But I think the employment debate needs to look beyond the employee count. The important question is how much human work the business purchases, from whom, and whether those purchases can support careers.

The humans outside payroll

Stripe's October 21 announcement describes Kubrio founder Vlad Stan as the education platform's only employee. It also says he plans to have an agent prepare payments to Kubrio's contractors and creators, with every payment subject to his approval. That payment workflow was a plan, not a reported completed deployment. [2]

The contractors matter. A payment prepared by software can still compensate a person. An employee count therefore misses part of the human contribution described in the company's own account.

Stripe supplies no contributor count, working hours or compensation figures. Its announcement is a promotional account, not an independent audit of Kubrio's labor needs. [2]

I see room for both opportunity and pressure here. A small payroll might coexist with substantial outside commissions, or with only occasional assignments. Calling either arrangement a one-person business tells a job seeker too little.

A solo founder can still have a team

Stripe Atlas's May 28 analysis makes another distinction explicit: its solo founders are people who started businesses without cofounders. The reported 63% share concerned C corporations formed through Atlas so far in the second quarter of 2026. It did not measure the share operating without employees. [3]

Who founded a company, who works on its staff, and whose services it buys are different questions. Treating these categories as interchangeable can turn a business formation trend into a confident employment forecast without supplying the missing employment evidence.

I would want claims about businesses run by one person to account for regular outside contributors. People remain contributors even when their names appear on invoices instead of an employee list.

AI gets the first attempt

FreshBooks reports that 86% of surveyed solopreneurs and microbusiness owners try AI before hiring help for a task they cannot complete independently. Its page, updated September 15, describes an online Wakefield Research survey of 500 respondents conducted June 22–July 1. [4]

That is a stated purchasing preference. It does not establish whether the AI attempt succeeded, whether a professional was subsequently hired, or whether the task would otherwise have generated paid work. [4]

Still, I consider this the sharper warning for freelancers. A buyer might now approach a professional after producing a draft or attempting a setup themselves. The available assignment could become narrower, more difficult or less frequent. Those are plausible consequences, not measured outcomes of this survey.

There could also be new demand from founders able to start businesses they previously could not operate. I cannot tell from these sources whether additional buyers would outweigh lost commissions. Both claiming the work is gone and promising abundant specialist jobs would outrun the evidence.

Look for repeat commissions

For someone seeking work with these businesses, I would examine what they actually commission. Look for published briefs, repeat engagements or services they openly say they buy. A founder's complaint about unreliable output suggests a problem. Repeated purchases of help provide better evidence that someone can build a livelihood addressing it.

Consider a hypothetical designer. A business that repeatedly pays for coherent visual identities after experimenting with generated images offers a different prospect from one whose owner is satisfied with rough results. The existence of imperfect AI output alone cannot tell the designer which business will become a client.

That distinction should guide learning. I would prioritize skills tied to recurring paid requests over a generic promise that every founder will soon need an AI specialist. It is reasonable to investigate a possible market before committing substantial training time to it.

For people who want employee roles, I would also distinguish a business advertising a staff vacancy from one purchasing occasional help. A contract can be worthwhile on its own terms; it should not be mistaken for evidence of a future position.

Self-employment is no automatic escape

Stripe's own results complicate the idea that displaced workers can simply become solo founders. Among solo-founded Atlas startups, it reports that median revenue during the initial six months fell 23% in 2025 from the previous year, while revenue at the top decile rose 19%. [3]

These are platform-specific revenue comparisons, not measurements of profit, personal earnings or AI's causal effect. [3] The stronger performance at the top does not establish a dependable alternative to employment for a typical worker.

I welcome tools that let someone attempt a business previously beyond their reach. I object to turning that possibility into an obligation to become an entrepreneur. Operating a business and obtaining dependable paid work remain separate achievements.

AI could enable a wider network of independent businesses buying services from one another. It could also leave more people competing for scattered assignments. I see no evidence here that resolves which effect will dominate.

My warning is specific: a successful business with few employees may offer other people only intermittent work. That would be bad news for anyone needing continuing assignments. I would investigate what these businesses repeatedly pay people to do, rather than blame workers for failing to become the next exceptional founder.