The era of artificial intelligence (AI) experimentation is ending, and insurers are now being judged on their ability to scale AI into measurable business outcomes.
At Insurtech Insights Europe 2026 in London, senior technology leaders shared their candid views on where AI tools are delivering real returns and where the industry is still falling short.
When asked about the most successful AI implementations, the panellists pointed to practical, high-impact use cases already transforming operations.
Pravina Ladva, group chief digital & technology officer at Swiss Re, highlighted claims and underwriting as standout examples. “In commercial insurance, we handle over 40,000 claims a year. AI helps us process them faster, detect fraud, and pay claims quicker, which is the moment of truth for customers,” she said.
She also pointed to underwriting transformation, where AI tools allow “clients to access decisions in minutes instead of hours or days.”
Dr. Fabian Winter, chief data & AI officer at Munich Re, emphasized that value creation is often cumulative rather than driven by a single breakthrough. “I don’t think there’s one big win (but) it’s hundreds of use cases creating value, he said, pointing to underwriting and the use of unstructured data as key areas of return.
Zuanella noted that across the industry, many of the biggest gains have been internal rather than customer-facing, at least so far. “Across the industry, the biggest wins have been internal,” he said, “streamlining processes, improving productivity, and empowering employees.”
Many insurers have spent years experimenting with AI, but few have translated pilots into enterprise-wide capabilities.
Ladva acknowledged the value of experimentation but stressed that the industry must now evolve. “First of all, I think everybody has done so many pilots, and I don’t want to dismiss their value, because pilots are how we learn. If we don’t learn, how do we know what to scale?” she said. “Pilots are important, but we now need to shift gears.”
For Ladva, scale is not about the number of use cases or deployments, but about outcomes. “For us, scale means embedding this technology into our core processes and the way we do business," she said.
Winter agreed but also noted that piloting will not soon go away. He argued that scaling AI also involves improving quality, which requires iterative testing. “Technology is evolving, and we need to continuously test whether it meets our underwriting or business needs, and whether it creates real economic value,” he said.
Zuanella warned that many organizations remain stuck in a cycle of experimentation without impact. “The proliferation of pilots is something we’ve seen before in many transformations. We start excited, driven by technology, and ask, ‘What can we do with this?’,” he said. “Piloting is easy. Scaling requires commitment, leadership, and risk-taking. That’s what will differentiate organizations.”
While technology often dominates AI discussions, the panellists repeatedly emphasized that success depends more on data quality, governance, and people.
At Swiss Re, Ladva described a long-term investment strategy that laid the groundwork for current successes: “Over the past eight years, we’ve invested heavily in data: quality, storage, and accessibility. Without that, AI wouldn’t be possible,” she said.
Winter echoed this sentiment, highlighting the importance of bridging technical and domain expertise. Munich Re has invested heavily in training and upskilling, including building an AI academy to ensure actuaries and data scientists can work effectively together. “Only with that overlap can you successfully implement AI into processes,” Winter said.
Despite progress, legacy systems remain a persistent hurdle. Winter noted that “all companies with history face this challenge… data sits in non-AI-native systems,” though he stressed that these barriers can be managed.
“Legacy systems should not become an excuse," stressed Ladva. "We all have tech debt and hybrid environments; it’s the reality We must build on that foundation.” Instead, she said, governance is emerging as the more critical issue.
“You need to monitor models, test them, and ensure they behave correctly,” Ladva continued. “That builds trust. And trust is fundamental… our industry is built on it. If we lose trust, we lose everything.”
