We turn the documents you

already hold into answers

you can defend

Knowledge bases for health policy, clinical care and health economics. Every answer carries the source behind it.

See what we do

Cancer research,

in words a patient

can act on

Under a minute, at a reading level you set, with every sentence traceable to the paper behind it. Built with SundVision, validated with patients at the University of Southern Denmark.

Cancer patient empowerment

Fourteen countries,

one shape,

every cell still cited

Comparison stops being fourteen separate searches. In build behind two international health policy programmes, with the Copenhagen Institute for Futures Studies.

How we build knowledge bases

Clinical guidance where there

is no network and no power

A sealed, solar-powered case holding treatment guidance and the reasoning to apply it. Critical presentations go to a person, not to the device.

Inside FORTIAID

The evidence under the models

that move health budgets

A named analyst accepts every parameter, and every published figure traces back to the document it came from. The architecture behind the modelling programme of Movement Health Foundation, in Ghana and Peru.

AI for health economics

WHAT OMNIMEM DOES

Answers your organisation can defend, drawn from the documents you already hold

Every project starts in the same place. There is a body of material that already contains the answer, and a question nobody can answer from it quickly enough. Policy documents, clinical guidance, published research, national statistics, regulatory text.

The material is not the problem. Getting a reliable answer out of it is.

We build the knowledge base that makes that material answerable, then build the thing that puts the answer in front of whoever has to act on it. Sometimes that is a policy team, sometimes a patient, sometimes a clinician with no signal, sometimes an analyst who has to sign their name to a number.

Four applications follow. The engine underneath them is the same one.

FOUR WAYS IN

Where to start

Each of these is the same capability pointed at a different reader, in a setting with a different cost of being wrong. Follow whichever one describes your problem.

Knowledge bases

For research teams and policy programmes

Retrieval over a collection you choose, joined to a knowledge graph of how the things inside it relate, so fourteen countries can be compared in one shape rather than searched one at a time. Every cell keeps a link back to the document behind it.

Currently in build behind two international health policy programmes, with the Copenhagen Institute for Futures Studies.

How we build knowledge bases

Cancer research, explained

For oncology departments and trial units

Peer-reviewed oncology turned into language a patient can act on, in under a minute, at a reading level you set rather than one it guesses. Every sentence is matched back to the sentence in the paper that supports it, and a sentence that cannot be matched is removed rather than softened.

Built with SundVision and validated with patients at the University of Southern Denmark.

Cancer patient empowerment

FORTIAID

For clinical teams without infrastructure

A sealed, solar-powered case holding treatment guidance and the reasoning to apply it, answering clinical questions where there is no network and no mains power. A defined set of critical presentations is routed to human judgement rather than answered by the device.

Field validated on site, with independent clinical review by the University of Southern Denmark.

Inside FORTIAID

AI for health economics

For ministries, insurers and financiers

The evidence layer underneath the models that decide where a health budget goes. We assemble the parameters those models need, with a named analyst accepting every value and a chain from any published figure back to the document it came from.

The architecture behind the modelling programme of Movement Health Foundation, applied in Ghana and Peru.

AI for health economics

HOW WE BUILD

What is true of all four

The four applications look nothing like one another. The commitments underneath them are identical, and they are the reason this work survives contact with a ministry, a regulator or a clinician.

Grounded in a collection somebody chose
Answers come from material you selected deliberately, not from whatever a language model absorbed during training. What is in scope is a decision, and it is written down.
Every answer carries its source
The passage behind a statement travels with it, through plain-language drafts and comparison tables alike. An answer you cannot trace is an answer you cannot defend to the person who has to act on it.
It declines rather than guesses
A question the sources do not answer gets that answer. The failure we design against is not an unhelpful gap, it is a confident paragraph about something the evidence does not cover.
A human stays where the stakes require one
A named analyst accepting a parameter, a clinician approving a patient letter, a critical case routed to a person. Where that boundary sits is agreed in writing at the start rather than discovered later.
Deployed where your data has to live, and yours to leave with
European hosting, your own infrastructure, or a sealed device with no connection at all. The material and everything derived from it stay yours, exportable in a non-proprietary format with help migrating. That sits in the agreement rather than being offered as goodwill.

Start with the question, not the technology

The useful first conversation is about the question your organisation keeps answering by hand, and whether the material that would answer it already exists somewhere in your files. Usually it does, in six formats and four languages.

What we do