Own the model. Keep the data.
Inspect the reasoning.

This is the single public page where a decision-maker — a data protection officer, a headteacher, a governor, a SENCO, a parent — finds every safety, data and accessibility commitment in one place, stated plainly, with what's done kept separate from what's promised. It says what other companies hide, and commits to things you can check.

Three configurations, and which of our claims hold in each.

How Aevum runs →

Evaluating Aevum for a school? The documents a data protection officer asks for are collected on one page, with an honest status on each.

The procurement pack →

Aevum's tools keep your data in the building, not on someone else's server. The information they use stays with you — it is not sent to a cloud service, not pooled with other organisations, and not our business model. A fuller explanation lives in our note, Where your data lives, where your model learns.

“No training on your data” is not the same as “no training”, and we won't blur-sm the two. Aevum trains a shared base model on Dawn, the AIRR supercomputer at the University of Cambridge, on data we curated — before any customer ever sees it. In the building, that model can then adapt to how you use it, and that local learning stays local. What it learns is split deliberately: anything specific to a child lives only in an erasable store held by the school and keyed to your own MIS, and never enters the model's weights. The model's own adaptation trains only on de-identified teaching material, with no per-child rows. We have not yet built the view that would let a teacher read that store back, and that is the next thing we owe you here. And because a model that has learned from personal information is itself a kind of record, we treat it accordingly — with the access, retention and deletion obligations that follow.

Teacher Manager's marking component: the thirteen DfE standards, published with the gaps →

• —We don't rely on US hyperscaler infrastructure to train our models, or run on a Silicon Valley account.
• —We don't take your data to a cloud we control, pool it, or sell it.
• —We don't carry advertising or monetise attention — you are the customer, not the product.
• —We don't pretend “no training on your data” means “no training”.
• —We don't pre-announce launch dates, run countdowns, or email you every week to keep you “engaged”. When it's ready, you'll get one email, with the link. That's the promise.

Every capability claim we make will be backed by a measurement we publish — and we publish the benchmark protocol before we run the experiment, so the test can't quietly move to fit the result. Until a number exists, the honest status of a claim is “designed to, being tested”, and that is exactly what this page will say.

Aevum is a British company. Its tools are designed in the UK, its models are trained on UK national research infrastructure, and it is accountable to UK law, including UK data protection law. It is built by a founder with twenty years in classrooms — first in Portugal, then in England — and designed by a neurodivergent founder with lived experience of disability, so accessibility is a first-principles design question here, not a nice-to-have.

Questions about anything on this page: hello@aevumfoundry.ai. This page is versioned; we update it as work completes.