Skip to content

Research · TUM Spin-off

TUM Spin-off. Built for industry.

demi is a spin-off of the Technical University of Munich, founded by three of its graduates. We research what makes AI reliable inside a company: connected data, permissions for people and agents, and results you can recompute.

  • Distributed systems
  • Artificial intelligence
  • Information systems
  • Information theory

Our foundation

Four research fields. One platform.

demi builds on these fields. Each one solves part of the task of bringing AI to a company's real data.

  • 01

    Distributed systems

    Many systems, one consistent state. Sync, conflicts and outages, without stopping any source system.

  • 02

    Artificial intelligence

    Models understand the question and explain the answer. Agents act, but only within their permissions.

  • 03

    Information systems

    IT creates value when processes and the organisation change with it. The org chart becomes the permission model.

  • 04

    Information theory

    How much does a record really know? Uncertainty, redundancy and noise become measurable.

Research areas

Five questions we work on.

Each area shows a research direction: the question, why it is hard, our approach and the open questions.

  1. 01Data layer and entity resolution
  2. 02Permissions for people and AI agents
  3. 03Deterministic agents and evaluation
  4. 04Distributed synchronisation
  5. 05Data quality through the lens of information theory
01 · Data LayerResearch direction

Data layer and entity resolution

When are three entries in three systems the same thing?

Why it is hard

The same part has one name in the ERP, another in the PLM and a third in the data sheet. Spellings, revisions and units differ. Merge the wrong records and every later number is wrong.

Our approach

  • Every value is a statement with its source, time and confidence.
  • A merge is a reversible link, never an overwrite.
  • Conflicts between sources stay visible.

Open questions

  • How confident must an automatic merge be before a person has to confirm it?
  • How does a store of documents become a graph with typed relations?

Diagram

Three systems

  • ERPItem 200-17 · Pump P200
  • PLMDrawing P-200, rev. C
  • Data sheet (PDF)P 200 · 7.5 kW

One record

Pump P-200
Item number
200-17ERP
Revision
CPLM
Power
7.5 kWData sheet, p. 2
Example: three entries, one record, every value with its source.
02 · Permission LayerResearch direction

Permissions for people and AI agents

How do you make sure an agent may never do more than the person it works for?

Why it is hard

Permissions are spread across the ERP, the file store and the org chart. An agent running on a service account quickly sees more than any single employee.

Our approach

  • Permissions come from your org chart, and demi never widens them on its own.
  • An agent may never do more than the person it works for.
  • Every decision is made per request and logged.

Open questions

  • How do you check automatically that the platform never shows more than the source system?
  • How do you explain clearly why someone sees something, or does not?

Diagram

  • Supplier prices

    • Anna, purchasingallowed
    • Agent mayallowed
  • Open purchase orders

    • Anna, purchasingallowed
    • Agent mayallowed
  • Personnel files

    • Anna, purchasingnot allowed
    • Agent maynot allowed

Where permissions come from

  1. 1Source system
  2. 2Org chart
The agent may never do more than Anna.
03 · Intelligence LayerResearch direction

Deterministic agents and evaluation

How do you get numbers from a language model that you can rely on?

Why it is hard

Language models write well but do not calculate reliably. The same question can produce two different answers. In quotes, orders and reports that is not acceptable.

Our approach

  • The model understands your question, the calculation runs in code.
  • The calculation runs deterministically on the records: same question, same result. The reading of the question can be wrong.
  • The model explains the result and names its sources.

Open questions

  • How do you measure whether the AI found the right thing? A set of golden questions per industry.
  • From what level of risk must a person approve, and when does approval become a formality?

Diagram

What is the open order value of customer 4711?

  1. 1Question
  2. 2Model understands
  3. 3Computed exactly
  4. 4Answer with sources

Check

The same question asked twice gives the same result. Anything else is a bug.

The model writes, the code calculates.
04 · Data LayerResearch direction

Distributed synchronisation

How do you keep ten systems in one state without stopping any of them?

Why it is hard

Source systems change all the time, go down or deliver late. Replacing core systems is expensive and risky. So the platform has to run next to them, not in their place.

Our approach

  • Read first: new connections start read-only.
  • Snapshots with a watermark. Changes follow wherever the system provides them.
  • Writing runs separately: as a proposal, with approval, reversible.

Open questions

  • When does a process on the platform match the legacy system, and how do you prove it in a parallel run?
  • How do you resolve conflicts when two systems change the same value at the same time?

Diagram

Read, first

  • ERP
  • CRM
  • Files
Snapshot with a watermarkData Layer

Write, only with approval

  1. Proposal as a change
  2. Approval by a person
  3. Write back
  4. Undo in one step
The source systems keep running. Nothing is replaced before it has proven itself.
05 · Data qualityResearch direction

Data quality through the lens of information theory

How much does a record really know, and for how long is it still true?

Why it is hard

Quality is usually measured as a single score. But a record can be complete and still outdated, or duplicated and still contradictory.

Our approach

  • Information as reduced uncertainty: which source tells us most about a value?
  • Use redundancy between sources as evidence, not as ballast.
  • Measure quality in several dimensions: correct, current, connected, readable for AI.

Open questions

  • What is the half-life of knowledge about customers, prices and contacts?
  • Beyond what input quality does the AI's answer stop getting better?

Diagram

Probability that the value is still right

Time since the last confirmation

  • Address
  • Contact person
  • Price
Schematic illustration, no measured values.

Publications

  • Published 2018 · Book chapter

    Soto Setzke, D.; Hoberg, P.; Murgoci, A.; Franzbonenkamp, S.; Gaß, J.; Wolff, T.; Krcmar, H. (2018): Digitale Transformation bei den Berliner Philharmonikern. In: Informationsmanagement und digitale Transformation. Springer Fachmedien Wiesbaden.

    DOI 10.1007/978-3-658-22624-4_10

Network

Connected with university, the founder ecosystem and funding programmes.

demi is a TUM Spin-off and is supported by programmes in Bavaria, Germany and Europe.

demiMunich

Science

  • Technical University of MunichAll three founders studied at TUM.
  • TUM Venture LabsWhere demi was born.
  • TUM studentsProject studies, interdisciplinary projects and theses on our research questions.

Founding and funding

  • UnternehmerTUMPart of the ecosystem, UnternehmerTUM Incubator programme.
  • EXIST-ForschungstransferBMWE programme, co-financed by the European Union (ESF Plus).
  • Federal Ministry for Economic Affairs and EnergyFunds demi through the EXIST-Forschungstransfer programme.

Programmes

  • ESA BIC BavariaBusiness Incubation Centre of the European Space Agency.
  • NVIDIA InceptionProgramme for AI start-ups.
  • AI5innovationElevator Ignite accelerator of the EU co-funded European Digital Innovation Hub.

Research with us

Your thesis on a real question.

We supervise master's theses, interdisciplinary projects and working students. Every topic is a real piece of our roadmap, with a contact person in the team.

  • From tree to graph

    Propose typed relations between documents, a person confirms them. Precision and recall are measured per relation type.

    • Master's thesis
    • IDP
  • Did the AI find the right thing?

    An evaluation framework with golden questions that measures full-text, fuzzy, graph and metadata search against each other.

    • Master's thesis
    • IDP
  • The half-life of business knowledge

    Track statements from public registers over time and estimate a decay curve per type of statement.

    • Master's thesis
  • How good must the input be?

    Measure the signal quality of documents and images and test whether it predicts the quality of the answer.

    • IDP
    • Working student
  • AI security along the OWASP LLM Top 10

    Prompt injection, tool misuse, data leakage: a reusable red-team framework for agents and MCP servers.

    • Master's thesis
  • Proving tenant isolation

    Property and fuzz tests that try to get from one tenant into another.

    • Master's thesis

Your chair supervises the thesis at your university. We provide the topic, access and a contact person in the team.

Open roles and topics

Security architecture

How demi secures AI agents

Whom an agent acts for, what it may see and trigger, how companies stay separate and where data and models run: the principles, with sources.

Go to the security architecture

Research that reaches the shop floor.

In 20 minutes we show you how the data layer, permissions and AI work together on a case like yours.

Your configurator

No use cases yet

Your configurator is still empty

Add a use case from any card or case page and see what it brings per year, with the calculation. Free, no sign-up, no commitment: try it out and change any number.

Browse use cases