The short version
I just crossed about $1.5M in ARR as a company of one, with Heidi in development for under two months. The reason it works is the part nobody puts in the announcement: I refused to vibe code. A solo founder has no team to absorb the maintenance that vibe coding quietly assumes, so the only thing that survives without hands to keep it upright is structure. I built agentic engineering from the first line, the substrate of specs, evals, guardrails, and review, and the substrate does the carrying a team would otherwise do. Vibe coding is low CapEx and high OpEx, and a company of one cannot afford the OpEx. This is the vibe coding crossover point made personal: the company of one is only possible on the far side of it, where the structure you paid for early keeps the system standing without people.
I crossed about $1.5M in ARR last month. By myself. Heidi, the AI cortex behind Falkster, had been in development for under two months. One person, eleven lighthouse customers, no team.
Every announcement of a number like that leaves out the mechanism. Here is mine. I am a company of one because I refused to vibe code.
A company of one has no slack for OpEx
When Google's vibe coding whitepaper draws its cost curve, it frames the choice as CapEx versus OpEx. Vibe coding is low CapEx and high OpEx. You start shipping immediately and pay for it forever in maintenance. Agentic engineering is the reverse: a real upfront cost to build the substrate, then a low marginal cost per feature.
For a funded team, that tradeoff is a genuine debate, because a team can absorb OpEx. They can throw bodies at the regressions. A company of one cannot. I have no one to keep a vibe-coded system upright when it starts falling over, which means high OpEx is not a cost to me, it is a hard ceiling. The only operating model available to a solo founder is the one where the structure stands on its own.
Vibe coding quietly assumes a team will be there to pay down the maintenance. Remove the team, and the only thing that survives is structure.
The substrate does the carrying
What a team usually provides is not just code. It is coordination, review, testing, the catching of mistakes before they ship. When you are one person, you have to build that into the system instead of staffing it.
So that is where the first hours of Falkster went. Not into a demo. Into the substrate: specs that define what done means, evals that act as the spec, guardrails that constrain what the agents can do, and review built into generation instead of bolted on after. I treat prompts as versioned operations, not one-off magic. That is the high CapEx on the curve, and it is unglamorous. There was nothing to screenshot for days.
Then it compounded. Once the substrate existed, each new feature got cheap, because the structure caught the failures a vibe loop would have shipped straight to a customer. The agents do the building. The substrate does the carrying. That is the only reason one person can hold the work that used to need an org.
The company of one is not a productivity hack. It is an operating model that only works on the far side of the crossover point, where structure stands without people.
This is not a story about being a hero
I want to be careful here, because the survivor version of this story is dangerous, and I have written about exactly that trap. The headline, one person, $1.5M, two months, is the right tail. It is not a template you can copy by firing your team and prompting harder.
The repeatable part is not the solo heroics. It is the substrate discipline. The founders who try to run lean by vibe coding their way to scale are building the high-OpEx system with no one to pay the OpEx. They are not running a company of one, they are running a company of one that is about to need ten people to clean up after it. The discipline is what makes the headcount optional, and the discipline is just agentic engineering applied early. It is the same line I draw in the old PM versus product builder ledger, lived at the smallest possible scale.
It is also why I have no patience for the "I built it in a weekend" flex. Weekend speed is the cheap half of the curve. A company of one only survives on the expensive half, where the structure holds the system up so I do not have to.
Pick one thing to try this week
If you are running lean, or dreaming of running solo, look at the most important thing you have shipped and ask whether it could survive a month with no one maintaining it. If the honest answer is no, you do not have a company of one, you have a maintenance bill waiting for a team. Pick that one system and put the substrate under it, the eval, the guardrail, the review, before you add the next feature. The leverage was never in shipping faster. It was in building the thing that holds itself up.
Also on Medium
Full archive →Frequently asked
How can one person reach $1.5M ARR with no team?+
By moving the upfront investment from people to substrate. Instead of hiring a team to write and maintain code, I built the structure that lets agents do the building reliably: specs, evals, guardrails, and review. That structure is what a team usually provides through coordination, and once it exists, one person plus agents can carry the work that used to need a funded org. The company of one is not a productivity trick, it is an operating model that only works past the vibe coding crossover point.
Why does vibe coding not scale to a company of one?+
Because vibe coding is low CapEx and high OpEx, and a company of one has no OpEx to spare. The moment a vibe-coded system starts throwing regressions, it demands the one thing a solo founder cannot supply: more hands to keep it upright. Vibe coding quietly assumes a team will be there to pay down the maintenance. Remove the team, and the only thing that survives is structure, which is why agentic engineering is the precondition for running solo.
What is the substrate in agentic engineering?+
The substrate is the structure around the AI that makes its output reliable: specifications that define done, evals that test it, guardrails that constrain it, and review built into generation rather than bolted on after. It is the high upfront cost on the agentic engineering cost curve. Once it exists, the marginal cost of each new feature is low, because the substrate catches the failures a vibe loop would have shipped. The substrate does the carrying that a team would otherwise do.
Is a one-person billion-dollar company realistic?+
The direction is real and already showing up in solo founders reaching seven and eight figures of revenue, but it is conditional, not universal. It works when the domain tolerates fast iteration and the founder has built real substrate underneath the agents. It breaks in regulated or safety-critical domains and for anyone who mistook vibe coding for the operating model. The headline number is the right tail. The repeatable part is the substrate discipline, not the solo heroics.
Should solo founders avoid vibe coding entirely?+
No. Vibe coding is the right tool for prototypes and exploration, where you will throw the code away before maintenance matters. The discipline is deciding upfront whether a thing is a prototype or a product, vibe coding the prototype freely, and putting structure under anything you intend to keep. The failure mode for solo founders is letting a vibe-coded prototype graduate into production without ever paying the structure tax.
What does refusing to vibe code actually look like day to day?+
It looks like writing the eval before the feature, so done is a test the agent must pass. It looks like versioning prompts as operations instead of one-off magic. It looks like spending the first hours of a project on the substrate instead of the demo. It is slower to the first screenshot and far faster to the hundredth, which is the whole point of building past the crossover instead of in front of it.
Comments (0)
Sign in with LinkedIn to leave a comment.
Sign in with LinkedIn