Why I Open-Sourced the Way I Run My Companies

Epistemic status: This one is argument more than data. The precedents are sourced. The reasons are mine, stated with the self-interest left in, because a piece about transparency that hides its motives would be a strange way to start.

What’s the Angle

The question arrives in different packaging but it is always the same question. The polite version is, why do you share so much about how you run your companies? The direct version is, what’s the angle? A CEO publishing his operating principles, his numbers with sources, his predictions with dates attached, his mistakes on a public ledger. In an industry that treats playbooks like trade secrets, it reads as either naive or a setup.

It is neither, and I would rather answer it once, in full, on the record, than one comment at a time. So this field note is the answer. The short version is a chain with four links. Make the nurse’s life better, and the patient’s life gets better. I can only reach so many nurses. So I publish how we do it, so other companies can see that scale and treating people well fit on the same P&L. And if enough companies act on that, healthy life expectancy, the number this country is worst at, starts to move. That is the whole angle. The long version follows.

The Tech Logic

Open source is not a philosophy that won because it was nice. It won because it was correct about where value lives.

Linux was copyable by anyone on earth, and that is exactly why it beat proprietary systems that were better protected and worse. Being open is what made it inspected, trusted, improved, and eventually installed on most of the computers that matter. Linus’s law named the mechanism: given enough eyeballs, all bugs are shallow. Secrecy hides your bugs from your competitors and from yourself. Openness ships them to a million reviewers for free.

In 2014, Tesla open-sourced its patents, and the headline read like philanthropy. It wasn’t. Musk had concluded the patents were protecting nothing worth protecting. The moat was never the blueprints. It was the factory, the pace, the accumulated daily decisions of actually building the thing. Publishing the patents cost Tesla nothing it valued and bought it the entire conversation about electric cars.

The tech world runs on this logic now. Companies build in public, publish their engineering blogs, open their roadmaps, and give away the research their competitors would once have killed for, because they all learned the same lesson: when execution is the product, the playbook is the cheapest part of the system.

And this logic is not something I’m adopting late. I started in software. I was 22 with a business plan and a chess table for a desk, built SpartX through its acquisition, then helped build PokitDok through its acquisition, and only then spent two decades operating healthcare companies. Open source is the water I came from. So when I built CultureAI, the operating system my healthcare companies run on, the obvious question wasn’t whether to apply the industry’s native logic. It was noticing what the software actually encodes. It encodes a culture. So I open-sourced the culture.

The Lab Is the Repo

Look at what is actually published at this domain, because it maps cleaner than I expected when I stepped back from it.

The manifesto is the README: what the system believes and why it exists. And the README has a first line, because my culture is not an abstraction with a website. Inside my companies it is four words, in order: Kindness, Unity, Humility, Patience. The Culture Lab is the documentation for what those four words cost to keep, at scale, under pressure.

The field notes are the commit log: every meaningful change to how I think, dated and numbered, including the ones that revise earlier entries. The prediction ledger is the test suite, and it is allowed to fail in public. Every claim carries its sources, which are the dependencies. The audio is the documentation read aloud. And CultureAI itself is being built alongside it, principles published before features.

Nothing on the site is a press release about the operating system. It is the operating system, versioned in public. When I change my mind, the change ships as a numbered entry instead of a quiet edit. That is the difference between a content strategy and a repository.

And the repo ships with its benchmarks, because published code you can’t run is just prose.

I list them not as trophies but as test output. The oldest excuse in this industry is that culture and growth are a trade-off, that you can have the happy team or the steep curve but not both. These numbers exist to retire that excuse. The source code you are reading is the same code that produced them, running in production, under load.

Why Give It Away

The real reason first, then the mechanisms.

I believe culture is a clinical outcome multiplier. That is not a slogan, it is the finding. Team engagement predicts patient experience. Safety culture predicts error rates. An eight-year study of hospital systems found staff engagement predicting mortality. The evidence is laid out, sourced, in the manifesto, and it all points the same direction: the same clinicians, the same buildings, the same budgets produce measurably different patient outcomes depending on the culture they operate inside. Culture doesn’t add to clinical quality. It multiplies it.

Follow that to its conclusion and something uncomfortable falls out. If culture multiplies outcomes, then every organization running a broken culture is quietly producing worse care than its own people are capable of, and every improvement in culture practice, anywhere, is a patient outcome improvement I will never see or get credit for. My companies serve one region of one state. The multiplier applies to all of them. So the highest-leverage thing I can do with what we learn is not to guard it inside my walls. It is to publish it where ten thousand operators can run it on patients I will never meet. Somewhere there is a nurse in a living room, employed by a company I will never visit, and whether her organization runs on fear or on trust is going to shape what happens in that room. That is who the multiplier is for. That is why the Lab exists.

And I hold one vantage on this that most people writing about healthcare, and nearly everyone writing about first principles, do not. I work in hospice. Part of my company’s work is being present at the end of life, which means I get to learn, over and over, what people actually care about when the auditing is done. It is never the title. It is never the optimization. It is people, time, and whether there was kindness in the room. The end of life is the most honest first-principles review that exists, and I get briefed on its findings every week. That is where my conviction about healthy years comes from. It is not a spreadsheet conclusion. It is a deathbed one.

Which surfaces a conflict I would rather state than have discovered. My companies are paid, in part, for the years when people need us most. If this industry gets better at manufacturing healthy years, some of those years are subtracted from the demand for what I sell. I understand that math, and I am publishing anyway, because I did not get into end-of-life care to root against life. If the mission shrinks my market, the mission was the point.

Everything else is mechanism. And the mechanisms are real, so here they are with the self-interest left in.

Writing is the forcing function. I do not publish what I think. I find out what I think by being forced to write it where operators I respect will read it. Every fuzzy idea I have ever had died in the drafting of a field note, and what survived was sharper than what I walked in with. The Lab makes my thinking pass inspection before my company has to live with it. Linus’s law, applied to a CEO: given enough eyeballs, all bad ideas are shallow.

Public receipts force private honesty. The prediction ledger works on me exactly the way I claim culture metrics should work on an organization. A confidence number with a resolve date, published, makes it embarrassing to quietly become someone who was never wrong. Open-sourcing the operating system extends that discipline to everything: it is much harder to run a company that contradicts a playbook the whole industry can read. Publication is an accountability device wearing a generosity costume.

It is the best recruiting filter I have found. People who read the Lab and feel something arrive already aligned. People who read it and roll their eyes were never going to thrive here, and both of us just saved a year of finding that out the expensive way. The 98 percent team satisfaction number is downstream of a lot of things, but one of them is that nobody joins this company surprised by what it believes.

And I am paying down a debt. Every framework I operate with was a gift from someone who wrote it down. The five-step algorithm came from a biography anyone can buy. Charlie Munger spent fifty years giving away his mental models in talks anyone could read. Buffett has published his operating philosophy every year since 1965. David Senra built Founders on the premise that the best builders left the instructions lying around and almost nobody reads them. I read them, and I have run healthcare companies on the published thinking of people who never met me, some of whom are dead. The Lab is that same move, made from my chair, aimed at an industry where nobody with a rocket company is coming to help. If something we figured out spreads because I published it, that is not lost advantage. That is the multiplier, multiplying.

The Moat Objection

The obvious question: doesn’t publishing the playbook arm the competition?

Here is what twenty years of operating says. A competitor can clone a repo and have working software the same afternoon. A competitor can read every word of this Lab, adopt the vocabulary, laminate the principles, and have nothing, because the value was never in the words. It is in the ten thousand Tuesday moments where somebody tests whether you meant it. I wrote a whole field note about the day a standard held for someone I liked, and the day it held for a deadline nobody thought mattered. You cannot copy those days. You have to have them, one at a time, under pressure, when it costs you something.

That is the one place the software metaphor breaks, and it breaks in my favor. Code executes the same way on anyone’s machine. Culture only executes on the machine that built it. The playbook is the cheapest part of the system. The decade of behavior that makes it true is the expensive part. Anyone who can execute that decade doesn’t need my documents, and anyone who needs my documents can’t shortcut the decade. The moat survives publication untouched, and the publication buys everything openness bought Linux: trust, inspection, and the conversation.

What Stays Closed

Open source has a convention worth honoring: you name what is not in the repo.

Patient information is never in it, obviously and absolutely. Individual compensation is not in it. Deals in flight, partner terms, and anything that is someone else’s confidence rather than mine stays out. So does the operational machinery: the specific workflows, vendors, and automation targets inside my companies. The philosophy is open source. The factory floor is not. And my team’s private struggles are theirs. The Lab tells my stories and the company’s patterns, never any individual’s worst Tuesday.

The rest, the principles, the methods, the numbers with sources, the predictions with dates, the mistakes with names on the ledger, is open. The test is simple: if it would make us better by being examined, it belongs in public. If it would only make someone else exposed, it doesn’t.

The Operator Application

You do not need a website to open-source your operating system. Start with one page: what you actually believe about running your company, written so plainly that an employee could catch you violating it. Publish it where your whole team can see it, which for most operators means the next all-hands, not the internet. Add one standard you commit to keeping, and one prediction about your own business with a date on it. Then let people watch.

The point is not the audience. The point is what publication does to the author. The moment your operating system exists where others can check it against your behavior, you either become the operator who runs it or the one who quietly deletes the page. Both outcomes are information you needed.

Ways I’m Wrong

The failure mode of building in public has a name: performance. A Lab like this could drift from receipts into content, from operating system into brand, until the publishing becomes the product and the company becomes the prop. I have watched it happen to better writers than me. The guard is the ledger, which does not care about my brand, and the people at HealthView who experience the gap between what I publish and what I do every single day. They are the code review.

And the moat argument could simply be wrong at some margin. Maybe there is a competitor for whom one published method is the missing piece. If so, the piece was cheap and the mission is served either way. I have made colder bets with worse odds.

The build continues in public either way. The repo is at cultureai.com, in text and in audio. Fork what you can use.