“AI pays off when we redesign the process, not just automate it”: Opus’ DACH lead Xenia Schmahl (Sponsored)

Founded in 2021, AppliedAI builds Opus, the AI transformation platform that rewires how enterprises run and ties every change to the bottom line. 

Many companies have tested AI, but few have scaled it across their workflows and systems. Opus closes this gap. Opus starts with how a process really runs today, and then redesigns it for AI. The result is a workflow built for the real world, which allows AI to scale in daily operations. Every run of the workflow and every decision, whether made by an AI agent or a person, is recorded.

Over time, this creates a digital twin of the organisation’s processes. Companies can always show what happened, why it happened, when, and based on which data. Before a project starts, Opus calculates what the new process will deliver, and it measures the result after go-live. Opus is now launching in the DACH region. The focus is on insurance, healthcare and other regulated industries, where strict rules make reliable and traceable AI especially valuable.

Why is 2026 the right moment for European enterprises to move AI from pilots to production, and why is DACH both a demanding and promising market?

AI models are no longer the bottleneck. When pilots fail, the reasons are usually operational: old systems are hard to connect, results are not measured, and there is no governance. These are exactly the problems we built Opus to solve. Opus takes an existing process, redesigns it for AI, and runs it in production with full governance and a complete audit trail. Today, this works reliably in regulated environments. That is why we are bringing Opus to DACH now.

DACH is a demanding market. Regulation is strict, data must stay in the region, and buyers expect governance from day one. But this is also what makes the market so promising. Banks, insurers and healthcare providers here run large, document-heavy back offices. This is exactly where reliable, well-governed AI creates the most value. 

What most often prevents European companies from scaling AI beyond pilots, and how should they choose the processes and solutions most likely to deliver value?

I see three main reasons why pilots stall. First, they are built on assumptions, not on the real process, so the first exception breaks them. Second, integration and governance come too late. Third, nobody owns a clear target, and often there is no way to measure success at all.

Good starting points share a few traits. The work is high-volume and repetitive. It spans several systems and involves many documents and exceptions. And someone owns the result. But the most important criterion is the return on investment. Not every process that can use AI should use AI.

AI pays off when we redesign the process, not just automate it. That is why, at Opus, we start by mapping how a process really runs today and measuring its baseline: time, cost, quality and error rates. Opus draws on its Work Knowledge Graph, a proprietary model of over a million enterprise processes organised by industry and jurisdiction. Then we calculate the expected return of the redesigned process, ideally before anything is built or changed. Once the new workflow is live, we measure again. This way, the value is clear before the project starts and provable after it goes live.

Then start with one process, prove it on real data within weeks, and expand from there. Choosing well also means knowing where AI does not belong. If an existing system already handles a process end to end, leave it alone.

Beyond cost savings, how should companies measure the success of AI-driven automation?

Cost savings are easy to report, but they are rarely the biggest benefit. I recommend five measures:

• Capacity: Can the team handle more work without adding people?
• Quality: How accurate is each field, for each case type? Measure this on test data that was never used for training. A single “97% accurate” number hides too much.
• Exceptions: A high automation rate is not enough. A wrong result that slips through unnoticed is worse than one that is flagged.
• Cycle time: How long does a case take from start to finish? Most delays come from waiting for human review, not from the machine.
• Auditability: In regulated industries, full traceability is a result in itself.

One example from our work: with Opus, a financial regulator cut licensing reviews from weeks to minutes of structured preparation, and every step remained fully traceable.

With the EU AI Act and strict data-protection rules, how can regulated sectors such as insurance and healthcare turn compliance and governance into a competitive advantage?

Many companies see governance as a cost. I see it as the thing that lets you go live. In regulated industries, the key question is not “Can AI do this?” It is “Will this pass risk, legal and security review?”

Governance becomes an advantage when it is built in from the start. That means AI acts within clear limits, humans approve the critical steps, every action is logged, and sensitive data stays in the company’s environment. This is how we built Opus: governance is part of the foundation. Every run and every decision is recorded, and over time this builds a Digital Twin of the organisation: a living model of how the business runs and what each part of it is worth. Companies that add compliance later often cannot use AI in these processes at all.

One more point: credibility with risk teams comes from precision. Be clear about which rules apply to you, from the EU AI Act to BaFin or FINMA. And never claim more than you can prove.

As AI agents take on more complex workflows, how will the relationship between employees and AI evolve, and what role do leadership, trust, and culture play?

Nobody can say how work will look in ten years. But for the next few years, we have a clear view. AI will free up a large share of the time knowledge workers spend today on routine tasks like typing, searching and checking. We should not pretend otherwise.

The real question is what happens with that time. If companies redesign their processes now, they can create meaningful new responsibilities for these people. Employees can then focus on the work where their judgement counts and where they create real value: complex cases, difficult decisions and contact with customers.

But these new roles do not appear on their own. They must be made visible. When we map a process with Opus, we define exactly where AI does the work and where a human needs to decide. This makes the human role clear from the start, instead of leaving it to chance.

This is where leadership matters most. Leaders need to explain openly what will change, and show employees where their future role lies. Trust grows when people can see that AI works within clear limits, and when they help shape the new processes themselves. Culture follows when people experience that AI takes over routine work, not their responsibility.

Over the next three to five years, where can European businesses differentiate from the US and China in enterprise AI, and what should leaders do today to prepare?

Europe will probably not build the biggest AI models. But that is not where most value will come from. Europe’s strength is trusted, well-governed AI in regulated industries, built under its own rules and in its own languages. That is a real advantage, if leaders use it.

There are four things leaders should do now:

1. Start with reality. Map how your processes really run today, including all the exceptions and workarounds. The process in the handbook is rarely the process people actually follow. With Opus Discover, AI agents interview the people doing the work and map how the process really runs, including the exceptions and workarounds.
2. Define your KPIs before you start. Decide what should improve, for example cycle time, quality or capacity, and measure where you stand today. Without a baseline, you cannot prove success.
3. Think in processes, not in tools. Don’t buy a new AI tool for every task. Look for a platform that can run many processes in one place, with one standard for governance.
4. Plan the new roles early. Decide where people will add value once AI takes over routine work, and communicate it openly.

The winners will not be the companies with the biggest models. They will be the ones that turned AI into part of how they operate, while others were still running pilots.

The post “AI pays off when we redesign the process, not just automate it”: Opus’ DACH lead Xenia Schmahl (Sponsored) appeared first on EU-Startups.