A small note on what we are building, and why.

We are starting a company. There is no product yet. This note is the first thing we will publish under its name, and we want to put on the record what we mean to build, before anything about it is true.

The company is called Brightroom. The four of us are in St. Gallen. The thing we are building is a GMAT preparation platform — although calling it that already understates the ambition, and overstates the current state of the work.

Why now

In March, the Graduate Management Admission Council announced that the GMAT, after a long stretch of incremental updates, would be replaced. The new test goes live in November. It is shorter. The essay is gone. There is a new section called Data Insights. Candidates can flag and revisit questions inside a section, with a small budget of answer changes.

Most of the prep industry has spent the summer publishing explainer articles. We have spent the summer reading the official specification, twice, and asking ourselves a different question: what would a prep platform look like if the people building it took the change as a chance to start over, rather than as a content-revision task on a roadmap?

The gap we keep noticing

A serious candidate, on any of the existing platforms, will be told their score after each mock. They will be told which topics they got wrong. They will be told which videos to watch next. What they will not be told — not by any consumer product we have used — is where, specifically, their score is bleeding. Whether the missed points came from one weak sub-skill, from pacing drift across a section, or from a slow accumulation of small errors. The shape of the weakness.

That gap is not a content problem. It is a measurement problem. The industry, as we read it, has spent the last twenty years solving the content problem and almost none of the measurement one.

What we are going to try to do

We are going to build the measurement layer first. An engine that estimates ability after every response and selects the next item to learn the most about the candidate per minute of their time. A representation of mastery across the actual sub-skills the test rewards. Pacing as a signal, not a footnote.

Everything else — content, lessons, pricing, marketing — will be derived from that decision. Not built around it, decoratively, the way most prep products treat their "adaptive engine" as a feature on a list.

What we are not going to do

We say this here because we think the category looks the way it looks because of these decisions — and because we think the best way to commit to a different shape of company is to write down, before we have anything to lose, the things we will not do.

1. 01No 1,200-page textbook. Shelf weight is not a proxy for seriousness; it is a tax on the candidate.
2. 02No sales team pitching a guaranteed score over the phone. A score is not sold; it is earned.
3. 03No chatbot wearing a graduation cap. The interface is the teacher; mascots are the tell that the interface is not.
4. 04No "ten secrets" article, no webinar countdown timer, no stock photograph of a diverse group of students. The category looks the way it looks because of these decisions. We are starting with the opposite ones.

What we have not figured out yet

Almost everything. We do not have an item bank of the size the engine will need. We do not have a pricing model. We do not know what the right diagnostic length is, in minutes. We do not have a written guarantee yet — although it is obvious to us that one will be needed, eventually, and obvious to us that the form of it matters.

We will work on these in public, as much as we can without leaking anything sensitive. This newsroom is meant to be the place where that work shows up. Some weeks the post will be a research note, some weeks a release announcement, and on the weeks when nothing is worth saying, we will say nothing.

Who we are

Four people, for now. Two of us are co-founders — Nicola Seiler and Joel Schleuniger. Both ex-consultants, both candidates of the test we are now trying to build for. One engineer, who has worked on adaptive measurement before, in a different industry. One curriculum lead, who has spent the better part of a decade inside this category and has a clearer read on what the published competitors actually do than anyone we have met.

We expect to be six within a year. Possibly never more than that. Six people, organised around a single product, is the shape of company we are trying to be.

Why we are writing this down

A founding letter is supposed to inspire. We have read enough of them to recognise the standard moves, and we are going to skip them. The reader of this note — if there is one — is almost certainly a candidate looking at the page out of a skeptical curiosity, or a future colleague looking at the company we are starting and trying to decide whether it means anything.

To both readers: the work has not begun yet. By the time you read the second note in this newsroom, it will have begun. Track us by the work. We will be here every few weeks.