Research
Applied research, done together with your organisation
We work with companies, public institutions and universities on questions that need a verifiable answer: does this system work? who does it disadvantage? what happens when it gets things wrong? is the investment justified?
What applied research means here
Applied research starts from a real problem an organisation has and ends with a result the organisation can use in a decision: a study, an assessment, a methodology or a working prototype.
What separates it from ordinary consultancy is method. We state the question explicitly, choose and document how we work, show the data we rely on, and say clearly what we could not determine. A result we cannot justify does not go into the report.
What separates it from purely academic research is its destination. The result is written to be used by someone making a decision, not only read by other researchers — though when it makes sense and the partner agrees, results can be published.
What we can do together
Four kinds of collaboration, from a single assessment to a multi-year project consortium.
Independent assessment of an AI system
We examine a system already in use or about to go live: what data it uses, how well it behaves on the cases that matter, who it may disadvantage, how explainable its decisions are, and what happens when it errs.
What you get
- Assessment report with documented methodology
- Analysis of unequal effects across relevant groups
- Risk list, ordered by impact and likelihood
- Concrete recommendations, each with an effort estimate
Studies, analysis and fundamental research
For questions that do not yet have an answer in the literature or in practice: field analysis, comparative studies, research on the organisation's own data, syntheses of the current state of a subject.
What you get
- A study or research report
- Documented datasets and analysis code
- A presentation of the results to your team
- Optionally, a publishable version of the results
Experimental development and prototypes
We check whether an idea works before the organisation invests in a full implementation. We build the minimum demonstrator that answers \u201cdoes this work or not\u201d and test it on real data.
What you get
- A working prototype and its source code
- Technical report setting out the solution's limits
- Effort estimate for a production implementation
- A reasoned recommendation: continue, change course, or stop
Consortia and funded projects
We take part as a research partner in national and international projects, in consortia alongside universities, companies and public institutions — from building the proposal to delivering work packages.
What you get
- Contribution to writing the proposal
- Ownership of research work packages
- Reports and deliverables to the funder's requirements
- Dissemination of results
How a collaboration works
A short, predictable process with a clear exit at every step. You are not asked to commit before you know what you are getting.
1. First conversation
An hour about your problem. What we want to know is what decision you intend to make on the basis of the result — the research question follows from that. Free and without obligation.
2. Written proposal
You receive the question as we would frame it, the proposed method, the deliverables, the estimated duration and the cost. If we do not think we can answer the question usefully, we say so here, not three months in.
3. Collaboration agreement
We set out in writing the rights over the results, the treatment of data and confidential information, and whether and how anything may be published. Nothing is left implicit.
4. Delivery, with checkpoints
We work in stages, with a checkpoint at the end of each. If intermediate results show the direction is wrong, we change it or stop — rather than continuing because it was in the plan.
5. Handover
You receive the report, the data, the code and a presentation for your team. Negative or inconclusive results are reported as plainly as positive ones.
6. What comes next
We discuss whether the result opens a next stage, whether it is worth publishing, and whether it could fit a funded project.
Why an association rather than a consultancy
A few structural differences that matter precisely when the result has to be credible to a third party.
Independent conclusions
We are a non-profit association. We have no product to sell you at the end of the assessment, so the conclusion cannot be shaped by what we intend to propose next.
Documented method, not opinion
Every result comes with how it was obtained. It can be checked, challenged and reproduced by someone else — including an auditor or an authority.
Negative results are reported
If the data does not support the hypothesis, that is what we write. A report that says \u201cthis does not work\u201d on time is worth more than an optimistic one that lets you keep spending.
Access to academia
We work with researchers and universities, so we can bring specialist expertise to a specific question without you hiring anyone permanently.
Honest about limits
We state explicitly what we could not determine and why. A declared limitation is useful information; a hidden one becomes an expensive surprise.
Clear rights over results
Who owns what is agreed in writing before work starts, not at the end. Publication, where wanted, happens only with the partner's consent.
Who we work with
Our collaborations start from a real need, not from an available budget.
Companies
That want to verify before scaling a system, need an assessment they can show a client or an auditor, or have an idea that should be tested before investment.
Public institutions
Introducing or reviewing digital systems that affect citizens, and needing an independent analysis of effects, risks and alternatives.
Universities and institutes
For research collaborations, project consortia and bridges to organisations that can put the results to use.
Civil society organisations
That need to understand and, with technical grounding, question digital systems affecting their work or the people they serve.
Frequently asked questions
How long does a collaboration take?
It depends on the question. A focused assessment of one system can take a few weeks; a substantial study or a prototype, a few months; a consortium project, years. The estimate is part of the written proposal, before anything is signed.
Who owns the results?
It is agreed in the collaboration agreement, before work begins. Typically the partner owns results specific to their own system, while the association keeps the right to publish the methodology — but only with the partner's agreement and without confidential information.
What happens to our data?
We work on a minimisation principle: we ask only for the data without which the question cannot be answered. Where it is stored, who has access, for how long and when it is deleted are all set out in writing. Where possible we work on anonymised or synthetic data.
Can the results be published?
Yes, if both sides agree. Publication is often in the partner's interest: a public methodology is easier to defend to an auditor or a regulator than one nothing can be said about.
Do you work with small organisations?
Yes. Size is not a criterion; the relevance of the question is. A well-framed question from a team of five makes a better project than a vague one from a corporation.
What does it cost?
Cost is calculated per project, based on the estimated effort, and appears in the written proposal. As a non-profit association, income is used exclusively for our statutory purposes. For externally funded projects, the cost structure follows the funder's rules.
Have a question that deserves a verifiable answer?
Write us a short description. The first conversation is free and without obligation — and if we do not think we can help, we will say so at the start.
Propose a research projectYour project does not have to be finalised. We are happy to discuss early-stage ideas.