When machines understand their own mistakes

Imagine this: At this very moment, billions of data points are flowing through the world’s factories – every single second. Sensors register temperatures. Cameras capture products. Control systems log every action, every process step. A digital heartbeat of modern production.

And yet around 90% of this data remains unused. It's not stored, not analyzed, not understood. It’s a hidden treasure trove that goes untapped – what’s known as “dark data”: data that is collected but never evaluated and therefore generates no added value. In the worst case, it can become a hazard for the business due to cyberattacks. This is a monumental waste. Such unused data contains information and answers to questions that cost companies millions. Why is scrap generated? Where is energy going to waste unnoticed? What distinguishes mediocre from excellent quality?

This is where EthonAI comes in – a Swiss scale-up founded by Julian Senoner (CEO) and Bernhard Kratzwald (CTO) who came up with their vision while working on their doctorates at ETH Zurich. With their AI software platform, they have developed a digital tool that not only monitors manufacturing processes but can actually understand them. Their solution analyzes data streams in real time, identifies the root causes of errors, discovers optimization potential, suggests solutions and intervenes before small deviations turn into costly problems. When the idea for EthonAI emerged, the then ETH doctoral students worked closely with Siemens – a collaboration that ultimately led to the founding of their company. Today, EthonAI is a rapidly growing international scale-up with around 50 employees – and was honored with the “Inventor of the Year Award 2025” in the “Open Innovation” category. In order to meet growing global demand, the company recently opened another location in New York City – a strategic step toward worldwide scaling.

EthonAI Founder Dr. Julian Senoner (CEO) & Dr. Bernhard Kratzwald (CTO)

Digital root-cause analysis reimagined

EthonAI transforms unused production data into actionable knowledge – through an intelligent combination of four key technologies:

• Causal AI recognizes not only correlations, but also true cause-and-effect logic – based on a knowledge graph that intelligently links all relevant process variables.
• Anomaly detection reports deviations long before they escalate.
• Process mining shows how processes are actually running – not just how they are planned.
• Computer vision automatically identifies visual defects that even trained eyes might miss – and integrates seamlessly into existing Siemens camera systems and production infrastructures.

All these technologies interlock within the various workflows of the EthonAI platform. They enable data-driven root-cause analysis and process optimization, support continuous real-time production monitoring and automate visual quality control. Together, they form a system that thinks like a digital quality engineer – it observes, compares, learns and draws conclusions. Let’s look at an example: In a chocolate factory, the same quality problems always reoccur, such as broken pieces, deformations or pralines that are stuck together. Traditional systems only recognize that an error has occurred, while the solution from EthonAI also identifies the cause. Temperature fluctuations, material batches, machine settings – all factors are correlated weighted, and presented transparently.

The ingenious part: The analyses are accessible without any programming knowledge. Thanks to an intuitive no-code interface, engineers can independently investigate causes and initiate improvements. Visual explanation features – such as color-coded heat maps – further increase decisions transparency and promote trust between humans and machines. Implementation is remarkably fast: The software suite is ready for productive use in less than three weeks. EthonAI accompanies the process with a dedicated roll-out team that supports data integration, training and onboarding – on site or remotely.

From Siemens pilot project to global platform

The technology was initially tested in several Siemens factories. Today, it is already in productive use in eight Siemens factories and is currently being tested in 20 more.

In Buffalo Grove (Illinois, USA), EthonAI supports automated visual quality control – an example of how software reduces scrap and measurably increases efficiency.

EthonAI has long since moved beyond the pilot phase: The solution is used globally in industry and is part of the Siemens Xcelerator Marketplace. As a result, not only does Siemens itself benefit from EthonAI’s innovative solution, but it is also available to Siemens customers and the entire Xcelerator ecosystem.

Thanks to its modular architecture, the solution scales effortlessly across plants and continents and can be seamlessly embedded into existing Siemens technologies.

The impact is measurable:

• Up to 80% less scrap
• Over 5% more throughput
• More than USD 1 million savings per plant
• 25% increase in operator efficiency
• ROI between 5 and 10 times

Sustainable, efficient – and explainable

Less scrap means less material consumption, less energy and lower emissions, making it a prime example of Siemens’ commitment to designing technology with a purpose, an impact and a sustainable design.

“You need courageous decisions,” says Senoner. “Siemens was one of our first customers – and has made exactly that kind of decision.” Kratzwald adds, “For us, this is the best proof that technology unfolds its greatest impact when people are willing to trust it.”

EthonAI shows how industrial AI not only improves processes but changes perspectives – and thus helps shape an industry that has a future. Even international innovation initiatives recognize the platform. The World Economic Forum honored EthonAI as part of its MINDS (Meaningful, Intelligent, Novel and Deployable Solutions) program as an example of how artificial intelligence can solve real industrial challenges.

With the Siemens for Startups Initiative, Siemens is expanding its commitment to supporting innovative companies like EthonAI on their journey and facilitating their market entry.

EthonAI demonstrates how AI is becoming a partner in production – not a replacement for humans, but an augmentation. The result: Factories that operate smarter, more sustainably and more resiliently – and an industry that is not only more productive, but also shapes the future.