Episode #177
May 22, 2026
Leading Digital & AI Transformation in Global Insurance
with Dr. Thomas Kuhnt from HDI Global SE
SPONSORED BY
Dr. Thomas Kuhnt, Chief Information Officer and Chief Operating Officer at HDI Global SE, joins us to discuss leading digital and AI transformation in corporate and specialty insurance. Drawing on his experience scaling AI in underwriting and claims and modernizing global operations, he shares insights on agentic AI, building enterprise capabilities competitors can’t buy, and how insurers can align people, processes, and platforms to drive resilience and innovation.
All Episodes Available On
From Algebraic Number Theory to Insurance Operations: The Pattern Behind Everything
Dr. Thomas Kuhnt grew up in northern Germany with a restless curiosity — the kind that leads a child to mix chemicals in the kitchen and watch what happens. That curiosity eventually led him to mathematics at the University of Bonn, then to a Fulbright grant and a PhD in algebraic number theory at the University of Illinois Urbana-Champaign.
It was a degree he was proud, at the time, couldn’t be applied to anything practical. “I was proud that you can’t apply what I’m working on,” he says with a laugh. That turned out to be wrong — the geometry of numbers, the logic behind patterns, became the foundation for everything that followed: coding theory, cryptography, and ultimately, systems thinking applied to global insurance operations.
After finishing his PhD, Kuhnt joined McKinsey, initially as a stepping stone to prove he could operate outside academia. He stayed for 15 years, rising to Senior Partner and leading the European Insurance Practice. When he joined HDI Global in 2018, he came in as COO. One month later, after a major IT project failed and the supervisory board decided they needed one person accountable for both operations and technology, he became CIO too.
“I don’t want IT and business to blame each other for why a project failed. I want one person I can hold accountable.” — HDI Supervisory Board
That dual accountability — still relatively rare in the industry — turned out to be the right structure for the age of AI. When you own both the process and the technology, you can’t outsource the hard questions to the other department.
Agentic AI: Don’t Wait, Start Iterating
At Insurtech Insights London, Kuhnt gave two talks — one on agentic AI, one on a COO framework for global operations. The message running through both was consistent: don’t wait for perfect conditions. Start experimenting. Iterate. The technology is maturing faster than most enterprise strategy cycles can keep up with.
“This is not going to go away,” he told the room. “We need to figure out how to work with it, capture the opportunity, and iterate fast.”
At HDI, the first meaningful AI deployment wasn’t a flashy transformation initiative — it was email input management. Brokers send submissions via Outlook, often with dozens of attachments in formats that vary wildly. Rules-based triage handled clean cases, but exceptions piled up. Agentic AI changed that math significantly, improving the performance of input management and enabling better triage — so underwriters only see the submissions they should actually be evaluating.
“Nobody likes to open those emails, read them, and then continue with triage. Applying agentic AI, you can really improve performance significantly.” — Dr. Thomas Kuhnt
It’s an unglamorous use case. It’s also one of the most consequential — because a submission that doesn’t get triaged efficiently is a submission that often doesn’t get quoted at all.
The End of Backlogs — and What Comes After
One of the most energized exchanges in the episode comes when Kuhnt and host James Benham align on a shared conviction: AI isn’t the end of the underwriter or the adjuster. It’s the end of the backlog.
Kuhnt’s data point is stark: 70 to 80 percent of what an underwriter does on any given day has nothing to do with actual underwriting. It’s rekeying data, reading documents, running background research, processing attachments. The skilled work — the judgment, the risk assessment, the relationship — accounts for roughly 20 percent of their time.
“That’s not really underwriting,” Kuhnt says. “But this is the first technology where they say: can I get access? Can I get more? They’re excited about it.”
For Benham, it’s the first time in 25 years of building software for insurance companies that he doesn’t have to say no to things that are technically feasible. The capacity constraint is dissolving. The next question — and Kuhnt is direct about this — is which things are actually worth doing.
HDI is also navigating a demographic reality that shapes the urgency: roughly a third of their workforce will retire in the next ten years. The pipeline of new underwriters coming from universities is thin. AI isn’t replacing the workforce — it’s helping a smaller, more focused workforce cover more ground.
Business Leads Transformation — Not IT
Perhaps the most operationally significant insight in the episode is Kuhnt’s answer to the question of who should own AI transformation. His answer: the business — specifically, the underwriters and operations directors in local markets.
“We are putting the leadership of most initiatives into the hands of the business leads,” he says. “Underwriters in the countries are leading the AI transformation. Because we will figure out the tech. But process and people are much harder challenges.”
“Business becomes IT — and IT becomes business.” — Dr. Thomas Kuhnt
This is a structural inversion of how most insurance organizations approach technology transformation. Centers of excellence, digital teams, and IT departments typically define the mission and own the roadmap. Kuhnt’s framework puts the people closest to the problem — and most affected by the solution — in the driver’s seat.
He’s equally direct about the process question. Kuhnt doesn’t want his teams to take existing processes and simply automate three steps with AI. He wants them to rethink the processes entirely. “That’s not a tech challenge,” he says. “The first question should always be why are we doing this — and that’s an ops question, not a tech question.”
Key Takeaways
- Don’t Wait — Start Iterating. The organizations building AI capability now are creating competitive advantages that late movers cannot purchase. Even imperfect early experiments build the organizational muscle — data literacy, change tolerance, process insight — that late movers cannot replicate.
- Business Leads, Tech Follows. The most effective AI transformations put operations leads in the driver’s seat. IT solves technical problems; business defines the mission. Failure is a business failure, not a technical one — and that accountability matters.
- The End of Backlogs Is Within Reach. AI liberates skilled workers from the 70–80% of their day spent on manual, non-value-add tasks. The underwriting backlog and the claims backlog are not capacity problems — they are workflow problems. AI solves workflow problems.
- Process First, AI Second. Applying AI a a broken process accelerates the breaking. The right starting point is always an operational question: why does this process exist, and does it need to exist in its current form?
[YOUTUBE_EMBED_PLACEHOLDER — paste YouTube episode embed code here]