Something interesting is happening in business formation data, and it should probably change how your district talks about careers.
According to Stripe's analysis of startups formed through its Atlas platform, solo founders now account for 63 percent of new companies this quarter, an all-time high. The gap between the best and the rest is widening too. Four years ago, top solo founders earned about 34 times the median revenue in their first six months. Last year it was 61 times. And the top performers were twice as likely to be building with AI at the core, using it to do work that used to require a payroll: building, shipping, supporting customers, and iterating, alone.
This issue is about what that means for the kids in your buildings, especially the ones in towns where opportunity has always meant a moving truck. It is the most optimistic thing we have written in a while.
Some folks have reached out to us asking how to subscribe themselves or their colleagues to this newsletter. Click here to do so. We send a newsletter every two weeks about the latest happenings with K-12 and AI.
IN THIS ISSUE:
The one-person company is booming, and career pathways haven't noticed
The skill the solo economy pays for is independence, and AI only builds it when the design is deliberate
63%
Share of new startups formed through Stripe Atlas this quarter with exactly one founder, an all-time high. Source
61x
Revenue multiple separating top solo founders from the median in their first six months, up from 34x four years ago. Source
10
Countries the average top solo founder sells into within their first month. Geography has stopped mattering. Source
20
Studies, out of more than 800 reviewed by Stanford, that provide strong causal evidence about AI's impact in K-12. Source
17%
How much worse students scored on an unassisted exam after practicing with a standard chatbot, versus classmates who never had one. Source
The economics writer Derek Thompson profiled what the Stripe data looks like at street level. A former Amazon employee wanted to open a Thai restaurant with his wife. A consumer chatbot surfaced an obscure rule allowing families to sell food made in a home kitchen, and the couple used AI to build their website, manage inventory, track compliance, and analyze their customers. The restaurant now sells out so often they are planning a storefront. Thompson's larger point sits between the two loudest camps in the AI jobs debate: the evidence doesn't show employment collapsing, with prime-age employment near record highs. It shows a specific kind of work exploding: the one-person company.
Now, hold that up against how K-12 talks about careers. Nearly everything in the career readiness apparatus, the pathways, the counseling, the resume nights, assumes the graduate's destination is a job application. Entrepreneurship, where it exists at all, is usually a boutique elective for a handful of business-minded kids. The data says that assumption is aging fast. If one person with judgment and AI fluency can do what used to take a team, then the founder's skills (spotting a problem, talking to customers, shipping, and adjusting) are becoming general-purpose graduate skills. Few districts teach them that way.
And notice one more Stripe number, because it is the one that matters most for most of America's districts. Top solo founders were selling into 10 countries in their first month, and about 40 by their second year. The customer has no idea what town the founder wakes up in, and definitely doesn't care.
OUR TWO CENTS
I grew up in a town of about 2,000 people. The deal in towns like mine was always understood: if you wanted to build something, you had to leave. Opportunity, staff, and funding were elsewhere, so our most ambitious kids became ambitious adults – just elsewhere, and every graduation was also a small goodbye.
That has changed. The availability of high-speed internet and, now, AI’s ability to help multiply the efforts of solo entrepreneurs have changed the paradigm. One person with persistence, judgment, and fluency with AI can build something serious from the same street their parents live on, selling to customers who don’t care about their zip code.
That is why I think these skills should be a graduation requirement, not an elective. Not because every student will start a company, but because in many small towns like Poplarville, Mississippi, for the first time in generations, there is a career path that doesn't require a moving truck. For small, rural, and even urban districts, AI fluency is not just college and career prep. It is probably the best local economic development strategy you have.
Teach a student to build where they live, and the kid may just stick around.
Move entrepreneurship from elective to pathway, with AI fluency built in from day one.
Let students run a venture, a service, or a storefront, with actual customers and actual revenue, however small.
Audit your career counseling for the employment-only assumption. Every pathway conversation should include building alongside applying.
Talk to your economic development office. A district that graduates builders is the best retention program a small town has.
Here's the catch in all this optimism, and it's worth being straight about. A founder is the ultimate unassisted exam. Nobody checks the work, nobody grades on effort, and the customer doesn't accept "the AI told me to." So the question beneath the company-of-one story is: are students developing the independent judgment that specific life demands? The real answer from the research: not necessarily or automatically.
Stanford's SCALE initiative reviewed more than 800 studies of AI in education and found only 20 with strong causal evidence, and not one high-quality causal study of student AI use in US K-12 classrooms. What the strongest available study shows should shape every tool decision you make. Wharton and Penn researchers ran a field experiment with nearly 1,000 high school math students at a school in Turkey, giving one group a standard chatbot during practice. Their practice scores jumped 48 percent. Then the AI went away for the exam, and they scored 17 percent worse than classmates who never had it. A second group used the same model wrapped in tutor-style guardrails, hints instead of answers, and teacher input built in. Their practice scores jumped 127 percent, and their exam scores held steady. Same AI. Different design. Opposite outcomes.
This surprised me: the students couldn't feel the difference. The paper notes their self-assessments were overly optimistic. They believed they were learning while the skill quietly failed to transfer. When independence is missing, it isn’t always obvious.
OUR TWO CENTS
None of this argues for less AI. The guardrailed version produced the biggest gains in the study and cost students nothing on the exam. It argues for two habits most districts haven't built yet. First, buy on design, not on brand. The Wharton study is the cleanest evidence anywhere that a tool built to coach beats the same model built to answer, so make "does it hint or does it solve" a standard question in every pilot. Second, measure with the AI off. Build regular unassisted checkpoints into any AI-supported course, problem set, cold draft, or oral explanation, because assisted performance flatters everyone, including (and especially) the student. If the skill is there, the checkpoint costs you a class period. If it isn't, you found out in October instead of at graduation. The kids headed for companies of one will thank you for the difference.
Add "hints or answers" to your tool evaluation rubric. Coaching designs and answer engines are different products wearing the same interface.
Schedule unassisted checkpoints in every AI-supported course, and tell students why: the goal is what they keep when the tool is off.
Ask vendors for evidence their tool improves unassisted performance, not just performance while connected.
Have students predict their unassisted score before a checkpoint, then compare. Closing that perception gap is a lesson in itself.
Jarvis left the corporate world to run a deliberately tiny business, and his argument has aged well into the AI era: growth is a choice, not a law, and a small operation built on judgment and craft can outlast a bigger one. For district leaders, it doubles as a field guide to the working life a growing share of your graduates will choose, and a useful provocation for anyone designing a pathway around it. Read it and then look at your CTE catalog.
Chat with ClassCloud
We’re listening. Let’s Talk!
This newsletter works best when it’s a conversation, not a broadcast. If you want to talk through how any of this applies to your district specifically—or if you have feedback on what would make this more helpful—just hit reply. We read and respond to everything.
Schedule a Virtual Meeting
Thanks for reading,
Russ Davis, Founder & CEO, ClassCloud ([email protected])
Tracy Walters, Director of Customer Growth ([email protected])
ClassCloud is an AI company, so naturally, we use AI to polish our content.
