you cannot be this person for yourself

I wrote a few days ago about the creation myth. The lie that great companies spring fully formed from someone's 3am epiphany. The reality being that they're chiseled. Iterated. Stood with. Notion in Kyoto for five years. Gamma eating a 95% drop-off for two. Clay at $0 revenue for five.

At the end of that piece, i talked about the pivot-vs-reduce-vs-remove-vs-delete distinction. That you're allowed to cut scope, kill features, delete flows, change the entire interface, but you're not allowed to pivot. Because a pivot is a reset. A pivot torches your compound curve.

Here's the part i didn't fully develop in that piece that i want to develop now: the overwhelming majority of pivots happen at 11pm on a Friday when someone is exhausted.

Not during strategy offsites. Not after rigorous analysis of the data. Not from a 6-month pattern of customer feedback. At 11pm. After a bad Tuesday. Because a competitor shipped something. Because a prospect said something mean on a call. Because the Product Hunt launch didn't go the way you thought. Because that thing your advisor said three weeks ago is still rattling around. Because you haven't slept right in a month.

And if you are alone at 11pm (no cofounder, no committed partner in the problem space) you are going to pivot. You might not do it that night. But the idea will plant itself. It will grow in the dark. You will convince yourself it was a rational decision by the time you announce it on Tuesday.

It wasn't. It was Friday 11pm. That's the whole thing.

Tyler said something to me a few weeks ago that stuck so hard i had to write it down.

He said: "what took me two years at my last startup, you and i are doing in two months."

Read that again. Two months. Two years. Same person.

His last startup was BloomText. A B2B SaaS company he ran alone-ish for years before Doozy. Real company. Real revenue. I'm not going to name the exact number but it's the kind of number where if you told a college student what it was, they'd tell all their friends.

And his honest, quiet assessment is that two months of us working together (two committed adults in the same problem space at the same time) is producing what used to take him two years. Ten times the velocity. Ten times.

Now, some of that is tools. AI is genuinely different now. Our codebase is 300,000 lines with zero engineers. I've written about the workflow and it blew up because people couldn't believe it. The tooling matters. Don't let anyone tell you it doesn't.

But the tools alone don't get you 10x. The tools are maybe a 3x. The other 3x is the cofounder. Specifically the sanity part of the cofounder, not the productivity part of the cofounder.

Which is what i want to name clearly, because i think everyone talks about this wrong.

When VCs tell you to find a cofounder, they frame it as a productivity multiplier. "Two people can ship faster than one. Division of labor. Complementary skill sets." You've heard this pitch. It's in every Y Combinator essay.

It's true. But it's not the main thing.

The main thing is this: a cofounder is a sanity multiplier. And sanity is the resource that actually runs out in year two and year three of building a company, not productivity.

Think about what's actually scarce in early-stage company building. It's not ideas. It's not energy, at least not at first. It's not even money, usually. There's more capital in the world than there is willingness to absorb it.

What's scarce is calibrated emotional regulation applied to strategic decisions over a multi-year timeline. The ability to show up on day 312 and evaluate a customer complaint with the same clarity you had on day 12. The ability to see the 2am panic for what it is, not react to it. The ability to stay excited about an idea for year three when the excitement has become routine. The ability to say "this feels bad but it's not actually bad, it's just Tuesday."

Most people can do this for 30 days. A smaller number can do it for 90. Almost nobody can do it solo for three years.

That's what a cofounder is for. They are your emotional regulation system. They are the kalman filter between your most panicked hour and your most critical decision. They are the person who catches the pivot before it becomes a pivot.

And THIS is the function that makes the "sell the house, stop the business" moment at 11pm get intercepted and defused into "let's talk about this Monday when we've both slept."

There is no productivity hack for this. There is no tool for this. There is no framework or journaling prompt or therapist session for this, because your therapist doesn't have equity in your company and doesn't know what you shipped Tuesday. They can't tell you if the panic is warranted. A cofounder can. That's the entire value proposition.

I want to tell you a story about two people from the University of Washington who figured this out before i did.

In 2008, a kid named Armon Dadgar was working on a research project at UW called the Seattle Project. Distributed systems work. Basically, a precursor to cloud computing, running background tasks across hundreds of thousands of donated laptops. Peak was around 200,000 machines.

His PI at one point handed him a task that he would later describe as "a complete nightmare": porting the sandbox to run on an early Windows Mobile phone. 16 megabytes of memory. Three-inch screen. A Microsoft attempt to enter the iPhone market that was, in his words, "horrific."

Armon spent a full quarter on this, and at the end, told his PI he would rage-quit if he had to keep going. His PI said, "ok, i've found another undergrad who doesn't know any better, hand it off to him."

The other undergrad was Mitchell Hashimoto.

Armon handed off the nightmare project and told Mitchell the truth: this is terrible, good luck. Mitchell (to his credit) made it about one quarter before rage-quitting himself. That was the start of their friendship. Two kids bonded by a bad research project.

They kept in touch. Senior year, they formed a five-person startup group called AMPED. Built some small things. A textbook comparison tool that hit $9k/quarter (good money for college kids, not real money). Some e-commerce experiments. Nothing that became a company. What they were really doing was training the muscle.

Then they graduated. Armon went to Kiip, a mobile advertising company, to work alongside Mitchell. (Mitchell recruited him so aggressively that Armon interviewed just to get him to stop.) Armon had actually deferred Berkeley grad school for a year to try it out. He never went back.

Two years at Kiip, they saw what a bad culture looked like. The CEO was, Armon said plainly, "a crazy person" who would tell the team they were going to be the next trillion-dollar company when Apple was worth $300B. Strategy changed every six months. The products that were working were ignored; the CEO's pet projects got funded. They watched the company ignore the data and follow the founder's ego into the ground.

Meanwhile, Mitchell had been doing a side project called Vagrant. It was originally a way to avoid wiping his laptop every few weeks at a consulting gig. Open source. Grew from nothing to hundreds of thousands of downloads.

At some point, Mitchell was working 8 hours at Kiip and 8 hours on Vagrant. He reached a breaking point. He left Kiip to start HashiCorp as, in Armon's words, "really just the Vagrant company."

Armon didn't join immediately. They were still best friends, lived three blocks apart in SF, saw each other five times a week. The conversations were about scope. "Is it just the Vagrant company, or is it something bigger?" They chatted about it for a full quarter. Just talked. Kept unwrapping the problem together.

Eventually they got to: "what if we tackle a bigger scope and take on a bigger chunk of infrastructure management?" Armon left Kiip and joined as cofounder. HashiCorp's mission became what it is today.

They ran HashiCorp for thirteen years.

Thirteen. Years.

They IPO'd in 2021. They sold to IBM for $6.4 billion in 2024. They shipped six major products over that span: Terraform, Vault, Consul, Nomad, Packer, Vagrant. They employed thousands of people. They defined a category. Armon is leaving in April 2026, and in his exit interview he said he's taking the summer off, going back to school to study architectural design, "just to use the left side of the brain for a change."

Thirteen years. Same person. Same cofounder. Same problem space. Wildly different product over time.

Here's the part that will stay with me forever. When Armon was asked about their disagreements over the thirteen years (you know, the famous cofounder breakup stories you hear at every YC event) he said this (quoting from my notes, we weren't supposed to record the talk, but i wrote this line down word-for-word because of how hard it hit):

"We really didn't have any major disagreements. Probably I mean, we had a few, but nothing that ever devolved into a shouting. It was always like, hey. Let's sit down and talk about our disagreement. Why do you feel the way you do? Why do I feel the way I do? And I think almost always what we realized is, we were coming into that decision with either different data points or different assumptions."

Read that. Slowly.

Thirteen years. Hundreds of consequential decisions. Billions of dollars on the line. Public company pressure. Private equity threats. IBM acquisition negotiations. Every single pivot-adjacent moment you can imagine.

And what he says is: we'd sit down. We'd ask why the other person felt the way they did. We'd realize we were operating on different data or different assumptions. We'd align on the data. Then the disagreement would usually dissolve.

That's not a productivity multiplier. That's a sanity multiplier. That's the 11pm test running for thirteen years in a row.

And there's a version of HashiCorp's story where Mitchell never meets Armon because their PI never hands off the Windows Mobile nightmare. There's a version where Armon goes to Berkeley in 2013 instead of joining Kiip. There's a version where the AMPED group never forms and both of them are working solo when Vagrant starts taking off, and Mitchell decides to just sell Vagrant to Red Hat in 2013 because he's tired and there's nobody telling him that Friday 11pm isn't the right time to make that call.

None of those versions produce a $6.4B company. The company is downstream of the friendship. The friendship is downstream of a rage-quit undergrad project in 2008.

You can't force that. But you can ABSOLUTELY look for it.

I was in the room with Armon recently. He came to speak at DubHacks Next at UW. They host a speaker series for the UW founder community. He walked us through the whole HashiCorp arc: the Seattle Project, Kiip, Vagrant, the six products, the IPO, the SaaSpocalypse, the private equity pressure, the IBM sale.

At the very end, in the Q&A, i got to ask the last question.

I asked him what he believed was the ONE thing that separated the founders who made it to the end from the ones who didn't. All the competitors they saw come and go. Docker, Mesosphere, CoreOS, the dozens of others. What was it?

He thought for a second. And then he said this (another line i scribbled into my notes app in real time because i didn't want to lose the phrasing):

"if I say what's the number one trait that defines the successful ones, it's not that they're the smartest, it's not that they're the hardest working, it's not that they're the most technical. They have very high grit, very high tolerance for bullshit over extended periods of time. Because it's a marathon. It's not a sprint. I know a lot of founders who go into it as a sprint. They're like, I work crazy hours and do all this stuff. But eighteen months, if they burn out, they hit eject. And you're like, this is not an eighteen month game. This is a very long term."

High tolerance for bullshit over extended periods of time.

Not intelligence. Not work ethic. Not technical skill. Not fundraising ability. Not even taste. Tolerance for bullshit over extended periods of time.

And here's what i want you to notice about his answer that he didn't say explicitly but is screaming through every word: you cannot sustain high tolerance for bullshit over extended periods of time alone. Humans are not built for it. Our emotional regulation decays when we're isolated. Our clarity erodes when the only voice in the room is our own.

Armon's answer is the sanity multiplier thesis in different words. The reason HE was able to last thirteen years wasn't because he was a superhuman with infinite grit. It was because when his grit ran out on a Tuesday, Mitchell's grit was still there. And when Mitchell's grit ran out on a Thursday, Armon's grit was still there. And when both ran out simultaneously, they had a process ("let's sit down, why do you feel the way you do, what data are you looking at") to get them through.

Grit is a renewable resource when there are two people. It is a depleting resource when there is one.

That's the whole game.

I keep coming back to this because i think it's something the AI-era founder ecosystem is getting wrong in a very specific way.

The narrative right now (and i hear it constantly in SF, in my group chats, on Twitter) is that AI is making solo founding viable again. That you don't need a cofounder because Cursor is your engineer and Claude is your marketer and Doozy is your EA and you can scale a billion-dollar company with one human and a stack of subscriptions.

This is technically true in terms of execution. You can ship a lot with AI. Tyler and i are doing it right now. 300k lines, no engineers, two humans.

But the execution part isn't the hard part anymore. I wrote about this. Context transfer is dead. Execution became cheap. Creation became cheap. A solo dev can ship more in a weekend than a team of five could ship in a quarter five years ago.

What didn't become cheap is emotional regulation over a multi-year timeline. And that's the thing that kills startups. Not lack of output. Not lack of features. Not lack of code commits. Founder burnout. Founder pivot panic. Founder decision fatigue. Founder ego that wouldn't let them kill the wrong feature. Founder depression in month 18 when the growth curve flattens. Founder isolation at 11pm on Friday.

AI can't regulate your emotions. AI can't tell you your panic isn't warranted. AI doesn't know if the product is dead or if it's just Tuesday. AI cannot be the person who says "we're going to talk about this on Monday after you've slept."

Only another human who has the same skin in the game as you can do that.

Which means in the AI era, the cofounder is MORE important, not less. Because when execution stops being the bottleneck, the bottleneck shifts to the thing execution was hiding: your ability to remain coherent over years.

You can now ship 10x faster. But you still need to stay sane for five to ten years. And sanity is the resource that gets expensive when everything else gets cheap.

Let me be honest about something.

I don't think there is a silver-bullet way to find a cofounder. I got mine the same way i've gotten almost everything that mattered in my life. Through a person who saw something in me before anyone else did.

His name is Caleb John.

I met Caleb during my DubHacks Next batch at UW. He was working out of Startup Hall, where we hosted our meetings. I was a first-gen kid with no real track record, no connections, no idea what i was doing. Caleb was one of the earliest people to advocate for me in the Seattle startup scene. He saw something in me when there was objectively nothing to see.

Caleb intro'd me to Jay Allen, which turned into my first real startup job at Moondream AI. A 4-person company that had just raised $4.5M. That job was the inflection point that made everything after it possible.

And a couple of years later, Caleb intro'd me to Tyler Brown.

Not at Foundations. Not at a conference. Not through a warm VC intro. Through Caleb. One of maybe four or five people in my life who have ever shown up for me the way he did.

So no, i don't have a "find your cofounder" playbook. What i have is: the right person brought the right person to me, and the product we've built together in six months is a thing i never could have built alone. Not in six months. Not in six years.

By the time Tyler and i started Doozy, the 11pm test had been tested a hundred times on small things. We knew what the other person did when they were tired. We knew how to pull each other out of spirals. None of that was luck. It was downstream of Caleb saying "you two should meet."

If i hadn't had Tyler, i don't know what i would have done. I don't think i would have started Doozy. Not because i couldn't ship a product alone. I've shipped four of them. Course Finder, Aristotle, the hackathon wins, the side projects. Those were all solo.

I wouldn't have started it because i know myself well enough now to know that i can't regulate solo over a five-to-ten-year horizon. Nobody can. Armon couldn't. Mitchell couldn't. Notion couldn't. Ivan Zhao and Simon Last went to Kyoto TOGETHER on a loan from Ivan's mom. They didn't each go to separate Kyotos. Clay couldn't. Kareem Amin built it with Varun Srinivasan. Gamma couldn't. Grant Lee built it with two cofounders. Slack couldn't. Stewart Butterfield built it with the same team that failed at Glitch.

Every single company i name-dropped in the Creation Myth piece has multiple cofounders. Every one. There is no outlier in the top-tier company list of the last decade that is a solo founder. Not one.

Go check. I'll wait.

So here's what i want to say directly to anyone reading this who is solo-founding right now.

I'm not telling you to quit. I'm not telling you your idea is wrong. I'm not telling you it can't be done. Technically, you can ship a lot, and you might make money, and some of you will get to a small exit doing it alone.

I'm telling you that the thing the ecosystem has not named for you clearly is what you're actually missing. You're not missing a second engineer. You're missing the 11pm test. You're missing the person who can say "the reason you want to pivot is that you're tired." You're missing the sanity multiplier.

And you're going to try to replace that function with other things. A coach. A therapist. A mastermind group. A Discord server. Twitter. Your partner. Your parents. Your dog.

None of those will work for the specific function i'm describing, because none of them have the context, the equity, and the inside view required to intercept a 2am pivot. They can help. They can't do THIS.

The only thing that can do this is another committed human in the same problem space as you. And i think if you're solo-founding and nothing else has clicked, the highest-leverage thing you can do this month is not shipping a new feature. It's finding that person.

Not a cofounder for the sake of a cofounder. Not someone to split equity with because YC said you need two on the application. Someone whose 11pm judgment you would trust more than your own, and who would trust yours in return.

That's the asset. Everything else is commodity.

Armon is taking the summer off. He's going to go do some traveling, take some architectural drafting classes for fun, and then figure out what's next. In the Q&A at DubHacks, someone asked him if he'd founded another company. He said yes, eventually. He doesn't know what. He said he likes the mission and the team-building and the process of it.

He did not say "i'm going to do it alone." He will find someone. He knows what the thirteen years with Mitchell taught him.

Me? I have Tyler. I'm going to be in this problem space for a decade. I don't know exactly what Doozy looks like in year three or year five or year ten. But i know it will be unrecognizable from today, and i know the mission will be the same: build a product whose UX compounds what a person does with AI in the right way.

And i know that in year six, when i'm exhausted and the growth curve flattens and some competitor ships something that makes me want to pivot, Tyler will tell me it's just Tuesday.

And i will tell him the same thing when his 11pm hits.

That's the whole game.

"They have very high grit, very high tolerance for bullshit over extended periods of time."

Yeah. But the bullshit only becomes tolerable when there's someone next to you tolerating it.

Find your person. Stay with them. Iterate. Reduce. Remove. Delete. Don't reset.

And when you want to pivot at 11pm on a Friday?

Go to bed instead.

-parsa