# The folder is the application

> Why we did not build another admissions portal, how a small school ended up publishing an open standard, and what we think happens when the application belongs to the family instead of the vendor.

- Published: 2026-11-17
- Author: Michael Janzen
- Audience: press, schools
- Canonical: https://admitprotocol.com/articles/the-folder-is-the-application
- Publisher: Admit Protocol, https://admitprotocol.com

Every school application system built in the last twenty-five years has the same shape. The school, or a vendor on its behalf, puts up a website. The family makes an account. The family types the student's name, date of birth, address and schooling into that website's fields, uploads documents into that website's storage, pastes essays into that website's text boxes, and clicks submit. Next school, next website, same typing.

The shape was forced by the technology of 1999. The only thing that could read a form was a person, so the person had to be put in front of a form. That is no longer true. The tools most families now have on their laptops and phones can read a folder of documents, understand what a school is asking for, and fill in the fields. The person is no longer needed at the form. So we asked what the application looks like when you take the form away.

## What we built instead

A family keeps the student's documents in one folder on their own computer: a note about the student, their answers to each school's questions, a report card, whatever else they have. A school publishes a short instruction file, the way a school once published a paper application, and an agent reads it: here is what we ask, here is how to send it, here are the rules. The family's agent does the carrying. A parent approves the result with a code. A person at the school reads it.

That is Admit Protocol: an open standard for the instruction file, the schema and the small set of operations an agent uses, and a hosted service that implements it so a school can be on it in an afternoon. Verde Valley School, where I work, is the first school on it. The specification is published so it will not be the last.

## Three decisions

**The student writes.** The obvious objection to AI anywhere near an application is the essay, so we put the rule at the center and enforced it in the server rather than asking nicely. An agent may organize, transcribe and send a student's writing; it may not draft, rewrite or polish it. Every piece of writing carries an attestation to that effect, confirmed by a parent, and a conforming server refuses writing without it. Schools are asked to read for thinking, interest and honesty, which is what they say they wanted all along. [The argument is here.](/articles/why-the-student-writes)

**People decide.** There is no scoring, ranking or fit model in the protocol, and the specification says there may not be. The industry pressure runs the other way; "AI-assisted review" is on every vendor's roadmap. We think the thing worth automating is the clerical part that was never meant to measure anyone, and the thing worth protecting is a human reader. A standard is a good place to write that down, because a standard is hard to quietly change.

**The family owns the application.** The folder never leaves the family's computer. The school gets only what the parent approved, under a published retention rule, and the service that holds it acts as the school's agent, not as a platform that owns the data. When a second school adopts the standard, the same folder applies there. The "common application" stops being something a vendor hosts and becomes something the family has.

## Why a school and not a startup

Because a school is the only party with nothing to gain from the form. Vendors are paid by the application or by the seat; the form is the product. Schools are paid by enrollment; the form is friction between a family and a decision. An admissions office that can read applications on a phone, get them from families who would have given up on the third website, and spend its time on people instead of data entry has every reason to want this and no reason to lock anyone into it.

It also means we have had to be honest about what is not finished. During the pilot, nothing is written into a school's student information system automatically; the admissions team copies a prepared block into Blackbaud by hand and marks it done. We say so on the schools page, in the agreement, and here. A connector exists and is being switched on first at our own school. We would rather a second school join knowing exactly what its Tuesday looks like than be surprised.

## What we think happens next

Not much, quickly, and then a great deal. A second school adopts the standard, and a family that applied to the first one finds the same folder works. An admissions software vendor implements the instruction file for the schools it already serves, because it is a week of work and its schools are asking. The Standard Application Online, which exists to let a family apply once to many schools, finds that the folder does that too, without the fee. Somewhere in there the question of AI and essays stops being a panic and becomes a rule everyone can point to.

Or none of that happens, and one school in Arizona has a better application process and a few hundred families who did not have to retype anything. That would also be fine. The specification is open; the cost of being wrong is small; the cost of waiting for a vendor to decide the shape of the next twenty-five years seemed larger.

## If you are writing about this

The facts are on [the press page](/press), including what we would rather you did not say: this is not AI applying to school for you, it is not AI reading applications, and it is not FERPA certified, because nothing is. Try it yourself with the live school on the front page; you can go as far as the parent code without sending anything. And if you run a school or build software for one and want to argue with any of this, hello@admitprotocol.com reaches the people who wrote it.
