Krones Brings Agentic Digital Twins to Beverage Production

Share this article
Share this article
Prioritise Us on Google
Krones uses digital twins powered by Microsoft, NVIDIA and more to automate operations. Credit: X/Krones
Built on OpenUSD, these virtual replicas allow autonomous agents to reason over data and execute decisions directly factory floors without human delay

Krones is rolling out agentic digital twins across its beverage production operations, a move the company says is improving efficiency and setting a new benchmark for digital transformation in manufacturing. 

Built with partners CADFEM, Microsoft, NVIDIA and SoftServe, the twins combine physically accurate real-time simulations with AI agents capable of reasoning, decision-making and automated action.

Rethinking bottling with real-time physics and AI

Bottling is a high-precision dance of ever-shifting variables, from filling speed and bottle geometry to pressure, flow and turbulence. 

These parameters must be monitored, predicted and optimised continuously to reduce waste and prevent downstream disruptions. 

Youtube Placeholder

Krones reports that, in just two months, its agentic digital twins have fundamentally reshaped machine and plant simulations, making it possible to model fluid behaviour during filling with high fidelity and interact with virtual replicas of bottling lines in real time.

From hours to minutes: simulation at production speed

Working with its partners, Krones has developed a physically accurate digital twin application that completes simulations in under five minutes – down from the three to four hours previously required. 

This acceleration allows engineers and operations teams to iterate, decide and optimise far closer to production pace, shrinking the gap between insight and implementation.

“With our trusted Microsoft Azure, Krones accelerates and scales its innovations through AI-powered tools and technologies,” says Gabriele Eder, General Manager Manufacturing at Microsoft Germany. 

Gabriele Eder, General Manager Manufacturing at Microsoft Germany. Credit: LinkedIn

“Together, we’re creating the solutions of today and the breakthroughs of tomorrow.”

Autonomous learning and closed-loop optimisation

Beyond speed, the system learns which configurations deliver the best outcomes on its own. 

Optimal solutions are pushed directly to physical machines, enabling real-time comparisons between virtual and actual performance and allowing instant adjustments on the line. 

The result is faster decision-making, smarter resource use and a tighter feedback loop between digital insight and operational execution.

“With real-time digital twins powered by OpenUSD, NVIDIA Omniverse libraries and accelerated computing, Krones is pioneering a new era of precision engineering,” says Naomi Betz, Senior Account Manager, Manufacturing & Industry (DACH) at NVIDIA

Naomi Betz, Senior Account Manager, Manufacturing & Industry (DACH) at NVIDIA. Credit: LinkedIn

OpenUSD – originally invented by Pixar – provides the foundational components for building shared, extensible 3D worlds that accelerate collaborative workflows and enable AI development in 3D.

A new standard for manufacturing

Krones positions this as a category shift for the sector. 

Traditional digital twins have typically required close collaboration with human experts, who interpret results and guide actions. 

By contrast, Krones’ agentic twins autonomously reason over real-time data, make decisions and trigger workflows automatically – moving from decision support to decision execution.

“With the Agentic Digital Twins, we are demonstrating that digitalisation and AI are not just visions for the future but deliver tangible efficiency gains and sustainable benefits today,” says Markus Tischer, Member of Krones’ Executive Board. 

“Together with our partners, we are setting new benchmarks for the beverage industry.”

Markus Tischer, Member of Krones' Executive Board. Credit: LinkedIn

As manufacturers race to modernise operations, Krones’s approach suggests a future in which physically accurate virtual replicas and AI-driven agents collaborate continuously with production lines – accelerating innovation cycles and translating digital insights into real-world performance at unprecedented speed.

Executives