As I write this, ROID has 1,575 people on it. By the time most of you read it, that number will have a 6 in the middle. You do not have to take my word for any of it — the count is live on the globe, pulled straight from the database, updated continuously. I did not round it up. I did not pick the flattering metric. It is 1,575, and in a few days it will be 1,600, and that is the whole of it.

I want to be careful here, because this is the kind of post where founders usually start lying. Not deliberately, most of the time. It just happens: the number gets softened into "nearly 2,000," the timeline gets compressed into something cleaner, the hard parts get sanded into an anecdote with a moral at the end. I would rather write the real one.

The Number, and What Is Actually Behind It

Of those 1,575 people, 490 have shared enough location data to place on a map. Those 490 sit in 31 countries, 366 cities, and 43 of the 50 US states. The largest single cluster is Irvine, California — 42 people — which is where this started, on a campus, with a product nobody had heard of and no reason to trust.

Forty-two people in one city is not a growth chart anybody would put in a pitch deck. But it is the number I think about most, because I know roughly how each of those forty-two arrived. They came from conversations. Someone showed someone else. A trainer put a program up and told their clients where to find it. That is a slower mechanism than paid acquisition and it produces a fundamentally different kind of user — one who was recommended, rather than targeted.

The spread from that one city to thirty-one countries happened without a marketing department, because there is no marketing department. There is no growth team, no engineering org, no one on call but me.

Building Alone

The honest description of solo founding is not the heroic one. It is that you are the constraint. Every part of the thing runs at exactly the speed of one person, and the parts do not wait their turn.

In a given week the work spans the iOS app, the web platform, the AI systems underneath both, the creator tools, the payment plumbing, the admin console, the support inbox, and whatever broke overnight. There is no handoff. When something fails at an awkward hour, there is no rotation to page — the rotation is one name long. When a decision is wrong, it is wrong in my handwriting, and I am the one who finds out, usually later than I should.

The part nobody warns you about is not the workload. It is the absence of a second opinion at the moment you need it. Every judgment call — what to build, what to cut, what a piece of feedback actually means, whether a quiet week is a signal or noise — gets made without anyone across the table to say I think you're reading that wrong. You learn to argue with yourself, badly, at length, and then ship something and find out.

I am not going to dress that up as character-building. Most of it was just hard, and some stretches were worse than hard. What I will say is that the thing survived it, and the reason it survived is sitting in the number at the top of this page: people kept showing up. Not many at once. Steadily.

What This Is Actually Built Against

Here is the part I care about most, and the reason ROID is shaped the way it is.

Every major social platform you use is optimized against a specific objective: time spent. That is not a conspiracy theory, it is just how the businesses are constructed. Attention is the inventory being sold, so the systems that decide what you see are tuned to produce more of it. Infinite scroll exists because a feed with a bottom is a feed you leave. Autoplay exists because a decision point is a chance to stop. The recommendation engine surfaces whatever holds you longest, and what holds you longest is frequently not what serves you.

The results of that optimization are now familiar enough to have picked up a nickname. People describe the state it leaves them in as brain rot — an hour gone, nothing gained, a vague worse feeling than they started with. That is not a failure of the technology. It is the technology functioning exactly as specified. The objective was engagement. Engagement was delivered.

Fitness content sits in a particularly ugly corner of this. The physique that performs best in a feed is frequently the one that is least achievable, least honest about its inputs, and most useful for making a viewer feel deficient — because deficiency is engaging. An algorithm optimizing for watch time will find that pattern on its own. Nobody has to be a villain for the outcome to be corrosive.

A Different Thing to Optimize For

So the design question I started from was not how do we build a fitness app with social features. Almost every tracker has bolted a feed on at some point, and almost every social network has bolted fitness on. Both of those produce the same object: an engagement product wearing a health costume.

The question was what a network looks like if the thing it optimizes for is whether you trained, ate, and recovered — and whether the people around you did too.

That changes concrete decisions. A feed built around what your actual friends are actually doing, rather than the most arresting stranger the model can find. AI that reads your own logged history and tells you something specific about your week, instead of generating another motivational rectangle. Creators who earn from the programs they build, so that the incentive to make something genuinely good outranks the incentive to be inflammatory. A design where the natural end of a session is I have what I came for, not an unbounded scroll.

None of that is charity, and I want to be precise about the business rather than vague about it. The app is free to download and use, with no subscription required. The AI runs on free credits that renew monthly, and heavy users can earn unlimited AI by inviting friends or take an optional plan. Creators earn from their programs. That is the entire model. Your attention is not the product being sold, because attention is not what the system is tuned to extract.

Across Generations

The assumption I brought into this — that it would be a college product, because it started at a university — turned out to be wrong, and being wrong about it has been the best part of the year.

It spread outward from campus in a way I did not engineer: to the parents of students, to trainers with client bases decades older, to people picking up resistance training for the first time in their fifties and sixties, and to people in thirty other countries who found it through someone they knew. Three hundred and sixty-six cities is not a marketing achievement. It is a word-of-mouth artifact, and word of mouth crosses generational lines far more readily than an ad campaign does.

What that taught me is that the appetite for a healthier relationship with a feed is not generational at all. Younger users are frequently the most articulate about what the current platforms are doing to them, because they have the most direct evidence. Older users often arrive having simply opted out of social media altogether, and are surprised to find a version of it they can tolerate. The complaint is shared. Only the vocabulary differs.

What Comes Next

I am not going to promise a number. Growth targets are the easiest thing in the world to write and the least meaningful, and I have watched enough companies chase one straight into building something worse.

What I will commit to is the shape of the thing. The feed stays organized around people you actually know. The AI keeps working from your real logged data rather than generic templates. Creators keep earning from work that helps somebody. And the moment a decision would trade a user's wellbeing for another few minutes of session time, it does not ship — which is a much easier promise to keep when there is no growth team whose quarter depends on breaking it.

To the 1,575 of you: you found this thing when it was rough, and a fair number of you told me exactly how rough, which is the most useful gift anyone has given this project. The bug reports, the feature requests, the messages that started with this is broken and here is how — those built the roadmap. Not a strategy document. You.

There is no team to thank here, which is the strange part of writing this. So I will thank the only group that made it work: the people who kept opening it.

ROID is free to download on the iOS App Store, with no subscription required. You can watch the community grow in real time on the globe, or see what creators are building.