for research groups and university services

Publications in the right place, without retyping.

New publications from your group reach the right researcher automatically, with a short plain-language summary and a link to the source. That summary only goes online once the researcher approves it.

Discuss your research system integration Start with a 1-hour Pitstop

No obligation · 30 minutes · online, with someone who builds, tests and fine-tunes it

Ethic is new. The team behind it isn’t. Ethic is the AI studio of Zenjoy, the digital product studio from Leuven that has been building and maintaining platforms, apps and integrations for more than 400 organisations since 2011.

See Zenjoy’s work

A researcher at his desk reads a publication that has landed in his tray straight from the bookcase.

At a glance

Works with: Your institution’s repository and your publication lists. Field data can be added later as an extension.
First: What your institution already offers centrally, and what your data management plan allows
Data: Personal data kept apart from research data, European hosting as standard

Honest advice

Sometimes an integration without AI is enough.

Your institution may already handle part of this centrally, and not every integration needs AI. We’ll tell you honestly which of the three fits, even if that means you don’t need us.

	What your institution already offers	An integration without AI	Your own AI workflow
Good for	Registering publications, open access and the link with ORCID. Your institution often handles that centrally already, through its own research repository.	Fetching publications, linking them to the right researcher and updating team lists. That doesn’t need AI.	Plain-language summaries, suggested keywords and domains, and organising field data (speech, photos, forms).
Watch out	You see your data the way the institution shows it, not tailored to your group, your website or your annual report.	A regular integration can’t organise free text, recordings or photos.	More work up front. You need a review step and agreements with your DPO.
What we do	First check what already exists centrally, so you don’t build anything twice.	Build, test on names that don’t match, and adjust when the API changes.	Build with an AI fact sheet, test with your own data and measure how often your team has to correct things.

Working mostly in Microsoft 365? Go to the Microsoft 365 integration

What you can do with it

Four uses around your research

Repositories and publication lists

Publications with the right researcher

New publications come in automatically and appear on the authors’ profiles and under the right research domain. Only doubtful cases still go to a colleague.

Abstracts and metadata

Summaries with sources

For each publication or project, the AI suggests a short plain-language summary, with a link to the source. The researcher approves or edits it. The assistant also answers questions about your own publications and protocols, with sources.

Go to Knowledge assistant

Projects and annual report

Reporting for your project

Publications, projects and figures from your systems come together in a first draft of your annual report or project report. Every number links to its source.

Go to Reporting

Extension · Field data

Recordings, photos and forms, organised

What participants or researchers collect in the field comes in organised, with the metadata your research needs. Doubtful cases go to a colleague.

Go to Forms and requests

How it works

A new publication, a summary with a source

This is how an integration with your research repository can work. It fetches and links on its own. The researcher approves what goes online first.

1

Fetch every night
The integration fetches your group’s new publications from your institution’s repository every night.
2

Match correctly
A fixed identifier, such as the staff number, matches each publication to the right researcher and domain. It does not match by name.
3

Summary with source
AI suggests a short plain-language summary, always linking to the publication in the repository.
4

You approve
The researcher approves, edits or rejects the draft. Nothing appears without approval.
5

Keep a record
The logs show what came in, which matches worked, what AI suggested and who approved what. This helps you spot errors.

Access and exceptions. The integration only reads, using its own access that your institution can revoke. Your IT department or library usually arranges access. If matching fails, the publication goes to a colleague.

AI fact sheet

Every integration gets an AI fact sheet

One page that says what the integration does, what it sees and who steps in. Your DPO, your IT department and your ethics committee can read along.

A man reads a page with a short, clear list.

See how an AI fact sheet works

Example

Publications from your repository, with a summary

Version 1.0 · October 2026

What it does

Every night, fetches your group’s new publications from your institution’s repository, links them to the right researcher and suggests a short plain-language summary for each publication. The website states that the text was prepared with AI.
What data

Publication data from the repository (title, authors, year, abstract) and the staff number for matching. No embargoed full texts, no research data. Exactly which fields your repository returns is settled at the start.
Model and location

A small language model that runs in the EU. The provider’s contract rules out using your data to train models.
Who checks

On its own: fetching publications and linking them to the right researcher. Via the researcher or a designated colleague: every summary, before it goes online.
What we keep

Which publications came in, which matches succeeded or failed, every suggestion and who approved or edited it. We keep this for 90 days.
Owner and stopping

Code, prompts and access are in the name of your group or institution. If your institution revokes the access, the integration stops.

Control

Agreements first, then AI

Some doors swing both ways, others only one. AI can do things on its own if they can be undone and stay inside your organisation. Anything that goes outside or decides about a person goes to a colleague first. So the integration can link a publication to the right researcher on its own. It can’t put a summary on your website.

Minimal permissions

The integration only reads what it needs, with its own access that your institution can revoke.

Personal data kept apart

If you survey participants, we keep their identity separate from the research data. The researcher sees pseudonymised data, not names.

Tested on what goes wrong

Names spelled differently, authors who have left, missing fields. We test with your own data before anyone relies on it.

Europe where possible

For each use, we choose a model and a location in Europe, and put both in the AI fact sheet. If that isn’t possible, we say so up front.

Who builds it

A new brand, built on fifteen years of craftsmanship

Ethic is Zenjoy’s AI studio, based in Leuven. The same team builds and maintains your workflow, and everything is in your name.

See Zenjoy’s work

15 years

building and maintaining software

400+

organisations

4 years

using AI daily in our own work

What Zenjoy built around researchers’ AI

Phonebox, KU Leuven (MeBioS)

The researchers built an AI model that recognises insects on sticky traps. Zenjoy built the product around it: a web application in the cloud, so the model became usable in the field.

See Phonebox on zenjoy.be

After the pilot

Read what happens after the pilot

We keep an eye on the integration

If your repository’s API changes, we adapt the integration.

Different model, same workflow

If a better or smaller model comes along, we switch without rebuilding the workflow. We do test it again.

Everything in your name

Code, prompts, documentation and access. Even if the researcher who set it up leaves.

Maintenance under a fixed agreement

Updates, monitoring and small changes are covered by a monthly agreement. New workflows are planned separately.

After a first conversation, you get a proposal with a clear scope and a fixed price.

Frequently asked questions

Questions from researchers, IT and DPOs

Question not listed? Just ask us.

A data processing agreement and security overview are available on request.

How much time does this take from our researchers?

Not much, but it has to come from the right people. At the start, you need someone who knows the task well, and someone who can arrange access to the repository or the API, often IT or the library. During the pilot of 4 to 6 weeks, your team approves the suggestions as part of the work it already does today. Every few weeks, we look at the figures together and adjust. What we need from you is set out in the proposal.

Where is our data stored, and does the model see our research data?

The model only sees the data listed in the AI fact sheet, for example publication data from your repository. Research data and participants’ personal data stay out, unless you explicitly agree otherwise with your DPO. As standard, the model and hosting are in Europe, and we name the provider and country in the AI fact sheet. If that isn’t possible, we say so up front. Your data isn’t used to train models.

Can the AI make mistakes? Who checks?

Yes, a language model can make mistakes. That’s why nothing goes online or into a report until a researcher or designated colleague approves it. Every suggestion links to its source, so you can check quickly. In the pilot, we measure how often your team has to correct things, and we agree in advance when that’s too often.

What if your repository or another system changes its API?

It happens. Zenjoy has been maintaining platforms and integrations for years after launch; an API that changes is part of that. Under the monthly maintenance agreement, we keep an eye on the integration and adapt it.

What does the AI Act mean here, and who is responsible for what?

An AI system that is developed and put into service specifically and solely for scientific research and development falls outside the AI Act (Article 2(6)). A workflow that supports your publications, website or administration usually doesn’t fall under that exception. Whoever uses such a workflow is the deployer under the AI Act and must, among other things, take measures to support the AI literacy of the staff who work with it (Article 4). The AI fact sheet states who has which role. This isn’t legal advice; your DPO or legal department has a say in it.

Does this fit our data management plan, and does the ethics committee need to see it?

That depends on the data and on your institution’s rules. If it’s only publication data that’s already in your repository, the question is usually simpler. If the integration processes participant data, recordings or fieldwork data, submit the AI fact sheet to your DPO and, if your protocol requires it, to the ethics committee. They decide what’s needed, not us. We adapt the integration to what your data management plan allows, not the other way round.

What should AI see in your research systems, and what shouldn’t it?

Show us what’s done by hand today and which systems your group uses. You get honest advice: what your institution already offers, an integration without AI or your own workflow.