.svg)
.svg)
.svg)
.svg)
This is a summary of an episode of Pioneers, an educational podcast on AI led by our founder. Join 3,600+ business leaders and AI enthusiasts and be the first to know when new episodes go live. Subscribe to our newsletter here.
TL;DR:
- Commercial fleet insurance remains unprofitable due to outdated pricing, static spreadsheets, and lack of personalized risk assessment.
- Telematics data and agentic AI offer carriers new ways to contextualize risk, improve underwriting, and price more accurately.
- AI agents can handle repetitive underwriting tasks while co-pilot models assist underwriters with context-rich decision support.
- Data quality and governance remain critical, as insurers struggle with legacy systems, master data issues, and fragmented pipelines.
- Startups face slow adoption cycles in insurance, but those aligning AI with business problems, not just hype, stand to create lasting value.
Before we dive into the key takeaways from this episode, be sure to catch the full episode here:
Meet Felix - Co-Founder of Draivn
Felix Kuhlmann, co-founder of Draivn, has spent two decades tackling the persistent challenges of commercial fleet insurance. From consulting to senior carrier roles, he has seen firsthand why this sector remains stubbornly unprofitable despite rich data sources.
At Draivn, Felix is working to turn telematics and contextual data into actionable insights that improve underwriting accuracy and risk selection. His approach is not just about better models but about reshaping outdated processes to align with dynamic fleet operations.
Felix also serves as an Insurance Executive-in-Residence at Accenture’s FinTech Innovation Lab, where he mentors startups and helps founders bridge the gap between insurance expertise and advanced technology.
A champion of domain-driven AI adoption, Felix emphasizes that real value emerges when data, governance, and context align. His mission: build risk intelligence that benefits insurers, brokers, and fleet operators alike.
Why Fleet Insurance Has Stayed Unprofitable
Felix Kuhlmann has studied commercial fleet insurance for over 20 years and describes it as “unprofitable for decades in every country I looked at.”
Traditional pricing relies on static spreadsheets, grouping vehicles into categories, and applying average rates. This outdated approach often fails to reflect real exposures, especially in dynamic operations where fleets expand seasonally or vehicles sit idle.
“Somebody is copying and pasting, and you just forget a few rows, and then you just have the wrong number,” Felix explains.
Despite the flaws, carriers continue offering the product because it is mandatory and premium revenue is attractive.
As Felix notes, “There’s always an idiot who gets it wrong, but that’s the one who wins,” reflecting the cycle of adverse selection.
The Promise of Telematics and Agentic AI
Kuhlmann highlights that the industry’s challenge is not a lack of data but how to use it effectively.
“There has to be a better way of using that data and bringing it to the process,” he says. Telematics, video, and contextual data provide rich insights into driver behavior and fleet performance.
Agentic AI lowers the cost of applying technology to niche areas like fleet insurance, where automation investment has historically lagged.
“Agentic AI allows you to deal with high variation in the process,” Felix explains, making it possible to tailor pricing and underwriting more closely to actual exposure. Success depends on building the right data pipelines, or as Felix puts it, “You need to figure out how you can actually put the right data context.”
Two Roles for AI Agents in Insurance
When asked how AI agents can be applied, Felix distinguishes between automation and augmentation.
“If the work is repetitive and rules-based, you probably can likely put an AI engine on top of it,” he notes, referring to repetitive, rules-based tasks that lend themselves to autonomous agents. The second role is more assistive, where AI acts as a co-pilot.
“How do you make underwriters stronger in applying the pieces and essentially just make them smarter?” Felix asks. In this scenario, agents enrich the context available to underwriters, making research and due diligence faster and more relevant.
As Felix emphasizes, consistency and calibration are key: “If you can get it to five to ten percent variation, I would bet that you can get any AI agent to the same.”
The Data Challenge in Insurance
Underlying all AI efforts is the question of data readiness. “It took us three years to just be able to frame this question: How many customers do you have?” Felix recalls from his time in a large carrier.
Master data management, customer hierarchies, and consent rules are often incomplete or inconsistent, making it hard to ground AI models in reliable context.
“If I don’t have a human sitting at the end who may catch something, I want to be even more sure that those pieces are watertight.” — Felix Kuhlmann
At Draivn, Felix focuses on building what he calls a “data factory” to normalize and structure inputs so they can feed both agentic AI tools and traditional actuarial models.
“Agents need clean, accessible data. And if you don’t have that, things kind of go sideways.” — Ankur Patel
Advice for Carriers and Fleets Embracing AI
Felix offers guidance to both sides of the insurance equation. For fleets, the first step is risk management.
“Assuming you want to have a decent price and good coverage, you need to be a better risk than average.” — Felix Kuhlmann
Telematics and AI tools can help coach drivers, track maintenance, and reduce claims, but fleets must also ensure insurers recognize these improvements. “Make sure you have a broker that can articulate that,” Felix advises.
For carriers, the priority is selecting partners willing to share data. “If a motor carrier doesn’t want to share data, you are an adverse selection if you are not,” he cautions.
By aligning risk behavior with transparent data sharing, both sides can achieve fairer pricing and better outcomes.
