Press · November 19, 2026

Our interview in Digital Engineering Magazin: probabilistic AI and AI agents

Digital Engineering Magazin has published an interview with Dr. Kevin Cremanns, co-founder of PI Probaligence, and Jason Easaw, who is responsible for technical key accounts. Under the headline "Ingenieure können sich künstliche Kollegen aufbauen" (engineers can build their own artificial colleagues), they discuss probabilistic AI for small datasets and AI agents that automate engineering workflows. It appeared on 18 November 2026, in German.

Visitors at the PI Probaligence booth watch an agent demo on the screen below the Probaligence logo while a camera operator films the scene.
The PI Probaligence booth at the CADFEM World Conference 2026 in Darmstadt.

Read the full interview (in German): digital-engineering-magazin.de

What the interview covers

The main topics, each with a link to more on this site:

  • Small data, local operation. Why STOCHOS is designed for limited, expensive datasets, and why it runs locally, so your data stays on your own infrastructure.
  • Simulation and measurement. How multi-fidelity modeling combines simulation results with a small number of measurements in one model.
  • Uncertainty. Predictions come with an uncertainty band. The engineer decides whether a result is enough or whether a simulation or an experiment is needed: uncertainty quantification does not replace validation.
  • Partners. Our work as part of CADFEM Group and as an Ansys Technology Partner.

Artificial colleagues: what the headline means

The "artificial colleagues" are AI agents: software that engineers instruct in plain language and that coordinates the steps of a workflow. In STOCHOS Flow, the Workflow Agent builds a workflow from a description, then runs it and fixes what fails.

The agent does not make the predictions. Those come from STOCHOS, our predictive engine, with an uncertainty band that shows where the model is reliable and where it is not. How the two fit together, agents that organize the work and a probabilistic model that delivers the numbers, is one of the interview's main topics.

The picture, also the opening photo of the interview, shows our booth at the CADFEM World Conference 2026 in Darmstadt, where we showed agent demos.

Next step
Product
STOCHOS

The predictive engine behind this work.

Product
STOCHOS Flow

Build, train, and deploy the workflow visually.

Interactive demo
Try to beat STOCHOS

Three use cases, about two minutes each.

See what STOCHOS can do with your data: request a demo. Releases and selected news by email: subscribe to the newsletter.

OlderWe're presenting at the European Coatings Show Confer...

← All news and guides

Partners, customers, and research collaborators
AnsysCADFEMSimuTech GroupMEScoTSNENAFEMS MemberBoschZFGEMUDLRAdler LackeMankiewiczDuluxPlixxentFraunhoferHochschule NiederrheinFUELL Lab AutomationHumotionUniversitaet HamburgRobert Bosch StiftungITficient