Education

Research Data Management Agent

A drive full of folders called things like messreihe_final_v3. Which one belongs to the publication?

Takes inventory of research data and proposes metadata and DMP building blocks.

Takes inventory of research data and proposes metadata and DMP building blocks.

Open agent

Enlarge demo

Public administration

Four processes that take up the most time today

Developed in Germany · Hosted in the EU

Developed in Germany · Hosted in the EU or on-premises

Delivered together with

Sovereign hosting in Germany, single-tenant from the first user.

Sovereign hosting in Germany, single-tenant from the first user.

Legal, tax and compliance expertise for legally sound deployment.

Legal, tax and compliance expertise for legally sound deployment.

Advisory and implementation for the public sector, starting with the AI readiness assessment.

Advisory and implementation for the public sector, starting with the AI readiness assessment.

An institute drive with 12,000 files and not a single line of description.

The measurement data sits where it was created, in each person's own folder on the institute drive. Exactly one person knew what a file name meant, and she is now at another university. When the funder asks for a data management plan or a journal asks for the dataset behind a figure, the search through the folder tree begins.

12,000

files

One institute drive holds six years of project work.

7

people

Each left behind their own folder names and filing system.

0

metadata records

No folder says where the data came from or what it was collected for.

2 days

per request

That is how long it takes to find the dataset behind a figure again.

The two days of searching per request are caused not by the number of files but by the fact that none of them is described; with metadata attached to each dataset, search time falls by 60%.

The 60% reduction in search time is this agent's headline figure. No institute drive keeps an inventory of itself, so the 12,000 files and seven filing systems are illustrative figures for six years of project work.

Three steps through the process.

The agent describes what sits on the drives and makes it findable; the project lead decides whether a dataset is published.

1.

1.

Take inventory of the holdings on the drive

The agent walks the folder tree and records file types, sizes, time periods and relationships. It lists identical datasets found in several folders as duplicates.

The agent walks the folder tree and records file types, sizes, time periods and relationships. It lists identical datasets found in several folders as duplicates.

Inventory for project FD-6284, 12,000 files, 613 marked as duplicates

2.

2.

Propose metadata for each dataset

For every dataset it recognises, the agent proposes a title, time period, method and authorship, and backs each proposal with a reference in the holdings. It leaves blank anything it cannot back up, with a note of what is missing.

For every dataset it recognises, the agent proposes a title, time period, method and authorship, and backs each proposal with a reference in the holdings. It leaves blank anything it cannot back up, with a note of what is missing.

57 datasets described, 6 fields per dataset, 11 fields without evidence

3.

3.

Enter datasets in the repository

The agent adds the described datasets, with their evidenced metadata, to the library repository as drafts. The project lead approves what becomes visible.

The agent adds the described datasets, with their evidenced metadata, to the library repository as drafts. The project lead approves what becomes visible.

57 entries in the repository, 12 linked to the data management plan

What the agent does not do.

It does not publish datasets.

It does not publish datasets.

It adds each one to the repository as a described draft. The project lead approves whether it becomes visible and under which licence.

It does not judge whether measurements are scientifically usable.

It does not judge whether measurements are scientifically usable.

It describes the period, method and scope of a series as the holdings document them. The research group judges the scientific quality.

It does not capture data on private computers or external hard drives.

It does not capture data on private computers or external hard drives.

Its inventory for FD-6284 covers the shared institute drive and nothing beyond it. Whatever sits on a former doctoral researcher's laptop does not appear in it.

Connected systems

Institute drive

library repository

data management plan tool

Operations

Deployment model

EU cloud or private cloud

EU cloud or private cloud

Data processing

GDPR-compliant, no model training on your data

GDPR-compliant, no model training on your data

Responsible teams

Research data management, institutes, library

Research data management, institutes, library

The questions that come up before the first meeting.

How quickly is the Research Data Management Agent ready to use?

Where is the data processed?

Where does the figure “60% less time spent searching” come from?

Related agents.

All agents

Platform

Industries

Services

Resources

Company

Pricing

From AI strategy to a running platform: we show you the way.

Talk to us

Log in

DE

EN

English

Platform

Industries

Services

Resources

Company

Pricing

From AI strategy to a running platform: we show you the way.

Talk to us

Log in

DE

EN

English
neuland.ai

The AI management and orchestration platform from Cologne.

Developed in Cologne. Operated in the EU or in your data center.

Certifications and compliance: ISO 27001, ISO 9001, GDPR, EU AI Act, DORA, StBerG, and BRAO
WirtschaftsWoche Best of Technology 2026 seal, 2nd place in the Process Management category for neuland.ai

© 2026 neuland AI AG