What is a company?

I have been asking myself this for weeks, and it started as a practical question rather than a philosophical one.

We are building Senpai’s operating system. Not a metaphor: the actual set of systems the company runs on, from how a lead arrives to how a member gets paid to what we publish. Every time I sit down to design a part of it, the same question shows up underneath. What is this thing I am building the operating system for?

It sounds like an easy question. It is not, and I know that because I have got it wrong before.

I had everything except a company

Between 2017 and 2021 I built Senpai into a community of more than two thousand people. Four cohorts. Five tracks. A newsletter reaching over three thousand. Our design course took a hundred and fifty seven students in its first cohort, forty five finished, and at least thirty of them now earn in dollars.

There was revenue. There were customers. There were outcomes I am still proud of. There was a brand people recognised.

There was no company.

I know that because of how it ended. Not with a bad quarter or a competitor or a failed product. It ended because I ran out of energy, and every single thing that made it work was running on my energy. I could not see who was struggling unless I was looking. Opportunities could not be routed unless I routed them. Rules could not be enforced unless I enforced them personally. The whole thing was a set of behaviours living inside one person’s head, and when that person stopped, so did the behaviours.

Everything I had was real. None of it was durable. And durability, I have come to think, is not one feature of a company. It is the whole definition.

What a company is not

Four things people usually mean when they say the word, and none of them are sufficient.

Not the people. People leave. Most of the people I taught in 2017 are doing better than me today, which is exactly what should happen. If the people are the company, then every good outcome is also a loss, and that cannot be right.

Not the product. Products change, sometimes completely. A company that cannot survive its product changing was never a company, it was a product with staff.

Not the legal entity. SENPAI FUTURES LTD, RC 9746107, is a real thing with real obligations, and registering it changed nothing about whether the work would survive me. The certificate is a container. It does not tell you whether anything is inside it.

Not the revenue. This one is the most seductive because it looks like proof. My first attempt made money. It was still not a company, because the money came from me personally doing the work, which is a job with extra steps.

What I think it actually is

The answer I keep arriving at is uncomfortable in how ordinary it sounds:

A company is the part of the work that keeps happening when the people who normally do it are not there.

Everything else is inputs. The people, the product, the entity, the money are all necessary and none of them are the thing. The thing is whatever persists.

And when you look at what actually persists, it is a short list.

Memory that is not in anyone’s head. In the old Senpai, if a founder asked me for a good JavaScript developer, I could answer for maybe twenty people, the ones I had personally watched deliver. For the several hundred others the honest answer was silence. That is not a knowledge problem, it is a storage problem. Founder memory holds about twenty people, decays with time, skews toward whoever you worked with most recently, and disappears entirely when you are unavailable. Institutional memory does none of those things.

Standards that survive being unpopular. A standard that only holds while someone is watching is a preference. It becomes a standard when there is a mechanism behind it: a required field, a step in a checklist, a review that cannot be skipped. Toyota did not ask people to care about quality. It put a rope on the assembly line that any worker could pull to stop production, and made pulling it the expected act rather than the brave one. The value did not live in anyone’s intentions. It lived in the rope.

Decisions with their reasoning attached. A decision recorded without the why is a rule nobody can safely change later, so it either gets followed blindly or thrown out entirely. Both are bad. Writing down why is what lets the next person disagree with you intelligently.

A record of what was actually done. Not reconstructed afterwards, which is always a story, but captured while it happened, which is evidence. This is the one I underrated most, and it is now the centre of everything we build.

Three tests

If you want to know whether you have a company or a group of people with a bank account, I think these are the questions.

One. Can it run for a week without you? Not survive. Run. Work gets allocated, decisions get made, standards get held, without you being reachable. Most founders discover the honest answer to this by getting ill.

Two. Would someone new know what good looks like here without asking? If the answer is no, then quality is your taste, and taste does not scale, cannot be delegated, and leaves when you do.

Three. If everyone walked out tomorrow, what would remain that was worth something? Not the furniture. The systems, the records, the relationships that belong to the organisation rather than to individuals, the reputation that can be checked rather than vouched for. If the answer is nothing, you have been running a very sophisticated freelance practice.

I would have failed all three in 2021. I would have argued about it at the time.

Which means a company is mostly built on purpose

This is the part that changed how I work.

You do not get a company by succeeding at something for long enough. I succeeded for four years and did not get one. You get a company by deliberately building the layer underneath the work, and that layer is invisible, unglamorous, and never urgent. Nobody has ever been forced to write down why a decision was made. There is no deadline for putting a value into a mechanism.

Which is exactly why most organisations do not have one, and why so many genuinely good ones quietly disappear when the person holding them together gets tired. It is not a talent problem. Nigeria does not have a talent problem. It is that the durable part gets built last, if at all, and by then the conditions that would have made it easy are gone.

So we are building it first, and in public

Senpai’s operating system is being designed before we scale rather than after, and we are documenting the whole thing as we go. These field notes are that documentation. Not a highlight reel published after it worked, but the decisions as they are made, including the ones that turn out to be wrong.

Two reasons.

The first is self-interested, and I would rather say so than pretend otherwise. We sell operating systems to other organisations. It would be absurd to sell that while running on WhatsApp and memory, and the fastest way to make sure we are not is to publish what we are doing where anyone can check.

The second is that this market has a trust problem and it is deserved. People here have been sold systems that were abandoned, hired firms that vanished, and watched work handed over as a repository nobody could run. Arguing against that is useless. The only thing that works is being checkable, over a long enough period that the record itself becomes the argument.

Where we honestly are

Senpai does not fully pass its own three tests yet.

The Collective platform is real, and it is the part I would point at: the record is built, the standards are enforced by mechanisms rather than by me, and it holds institutional memory that no longer depends on what I remember.

The rest is in progress. Our own client intake, our own evidence layer, our own delivery systems mostly do not exist yet. If I got hit by a bus this week, Senpai would not run for a month.

So the honest answer to the question in the title, for us, today, is: we are becoming one. And the point of writing this down now, before it is true, is that in two years you will be able to check whether we got there, using our own record, against our own definition.

That is more or less the whole idea.