Brokers dominate AI talks at ITC Barcelona
Brokers have relied on instincts honed over decades—strong relationships, deep market knowledge, and a preference for proven methods. At ITC Barcelona, Holm Schimanski, partner and chief AI officer at tigerlab, explained how those instincts may now lead to expensive technology mistakes.
Schimanski drew from recent experience. His firm completed a live rollout of a broker management system across multiple countries, uncovering patterns that repeatedly trip up brokers when purchasing software. The discussion avoided flashy AI demonstrations or abstract use cases, focusing instead on three recurring issues in broker technology decisions.
Trust in vendors can backfire
The insurance sector thrives on personal connections, and brokers often favor trusted vendors when selecting software. That trust sometimes proves misplaced. Schimanski cited cases where brokers committed to long-term contracts with familiar vendors, only to find the software lacked basic integrations or couldn’t scale beyond a single office. The damage extended beyond licensing fees, creating lost time and manual workarounds.
Related: PRA fines insurer £4.165m over faulty FSCS data
Customization creates long-term problems
Brokers often request software tailored to their exact workflows. Vendors, eager to secure deals, frequently agree. Schimanski warned that customization can become a liability. Each adjustment adds complexity, slowing future updates and increasing costs. Over time, the software grows so specialized that migration to a new system becomes nearly impossible without starting over.
Tigerlab’s rollout exposed another challenge: brokers rarely anticipate future needs. A system built for today’s workflows may struggle with tomorrow’s regulatory shifts or new product lines.
The insurance industry has long grappled with legacy systems customized to the point of unmanageability. The difference now is the speed at which these systems become outdated. AI and automation are accelerating change, making rigid, bespoke software riskier than ever.
AI should be built into the system, not added later
Most brokers Schimanski encountered wanted AI as an optional feature—a chatbot for customer service or a tool for automating routine tasks. While generative AI hasn’t introduced new types of fraud, it has made producing convincing fake evidence easier. A well-integrated system can flag these inconsistencies before they result in losses.
Related: Cost hinders digital trade for MGAs
Schimanski acknowledged the value of relationships and workflow-compatible software. However, he urged brokers to ask tougher questions before signing contracts. How difficult is it to switch vendors? What happens when regulations change? Does the system process data in real time, or does it rely on nightly updates? The answers often reveal more than any sales pitch.
Tigerlab’s founder and CEO, Tobias Bergmann, has spent 18 years developing insurance platforms. His perspective on long-term software was direct: “Most systems aren’t designed to last. They’re designed to sell.” The ones that endure, he noted, favor simplicity over features, flexibility over customization, and data over assumptions. AI may dominate current discussions, but the real test is whether a system remains functional when the next major shift arrives.
The commercial combined market has faced similar challenges, where outdated systems struggle to keep pace with evolving demands.
