When you think about AI in industry, it’s easy to imagine futuristic visions that feel years away from reality. But for Covestro, a global leader in high-tech polymer materials, that future is already here.
In the latest episode of Lemongrass Roots, Nils Janus, Chief AI Officer at Covestro, shares how the company has taken artificial intelligence from a series of small experiments to an enterprise-wide capability that’s reshaping everything from production lines to financial forecasting.
Starting Small: Building AI from the Ground Up
When Nils joined Covestro back in 2018, there was no grand AI strategy or roadmap — just a question: could data science and machine learning deliver real value in a highly regulated industry?
The first steps were grassroots. A small team of data scientists began running experiments and looking for willing collaborators across the business. Early wins came when machine learning models started optimising processes in Covestro’s chemical plants — demonstrating tangible results that plant managers could see in their own KPIs.
That, says Nils, was the turning point. “Once people saw the results in their own metrics, it kicked off an avalanche of demand across departments,” he explains. Suddenly, AI wasn’t a research project; it was a business enabler.
Turning Proof Points into Enterprise Strategy
As those early use cases began to prove measurable value, Covestro moved to scale. By 2023, the team was ready to take AI from experimentation to strategic execution.
They began engaging senior leaders through structured AI briefing sessions — four-hour one-on-one conversations with every board member and executive. The goal was to educate, spark ideas, and align on what AI could mean for their individual functions.
The result was a company-wide shift. Executives weren’t just endorsing AI from the top down; they were becoming champions within their own areas. “We made it personal,” says Nils. “Each leader had the chance to ask questions, voice concerns, and explore how AI could create value in their world.”
AI as a Colleague: The Covestro Virtual Assistant
One of Covestro’s boldest innovations is the Covestro Virtual Assistant — an AI-powered agent that acts as a digital colleague for thousands of employees.
Designed to simplify how people interact with enterprise systems like SAP, the assistant enables employees to simply express what they need (“create a purchase order,” for example) rather than navigating complex interfaces or remembering transaction codes.
Around 30% of Covestro’s white-collar workforce now use the assistant daily — a clear sign that it’s becoming part of the company’s digital DNA. “It’s a colleague you can ask for help,” says Nils. “Someone who just gets things done.”
The Next Frontier: An Agentic Future
Covestro’s AI vision goes beyond internal automation. Nils describes an “agentic” framework where Covestro’s own agents could one day collaborate with those of partners and software providers like Lemongrass.
By sharing interoperable AI agents, organisations could exchange expertise seamlessly, dramatically speeding up innovation and reducing costs. It’s a model that challenges traditional enterprise software relationships and opens the door to more collaborative, adaptive ecosystems.
Real-World Impact: From Production to Treasury
AI at Covestro isn’t just theoretical — it’s driving measurable results.
In 2024, the company launched the world’s first fully autonomous batch production line in a chemical plant. Multiple AI models work together to optimise production, detect when quality targets are reached, and even eliminate the need for certain physical testing — reducing cycle times and increasing throughput.
Elsewhere, in finance, an AI-driven liquidity forecasting model is helping Covestro predict cash demand with unprecedented accuracy. That’s allowed the company to lower liquidity lines with partner banks, cutting costs while maintaining the same risk profile — generating millions of euros in annual savings.
Advice for Enterprises Getting Started
For organisations still at the early stages of their AI journey, Nils offers two key pieces of advice:
- Experiment widely, fail fast. Don’t bet everything on one “sure thing.” Test multiple ideas, learn quickly, and double down on what delivers results.
- Build internal capability. AI is too core to your future operations to fully outsource. Even a small, trusted internal team can anchor your strategy and keep your transformation on course.
Leading by Example
Covestro’s story is a compelling blueprint for how large, complex enterprises can unlock the true potential of AI — not by chasing hype, but by combining experimentation, executive engagement, and measurable value.
As Nils puts it, the ultimate success of AI comes when it stops feeling like technology — and starts feeling like a trusted colleague.
Listen to the full episode of Lemongrass Roots featuring Covestro’s Chief AI Officer, Nils Janus, to hear more about the company’s enterprise AI journey and what’s next for intelligent automation in manufacturing.
