Robert Clark is the founder and CEO of Cloverleaf Analytics, a leading provider of insurance intelligence solutions.
Here is a scene that is playing out at insurers of all sizes in the industry: There are teams of underwriters that have AI-powered tools on their desktops, but they still rely on antiquated spreadsheets, flood maps and inspection reports. While some insurers have invested millions into machine learning and GenAI, when it matters most, the new tech looks more like a glorified calculator than a game-changer.
Most in our industry have been so focused on building smarter individual AI models that they’ve missed the bigger picture. The real breakthrough won’t come from a single super-system or model. It is more likely to come from swarms of AI agents (with human oversight)—specialists that debate, disagree and ultimately deliver decisions in minutes rather than days.
Where AI Falls Short Today For Insurers
Here’s the uncomfortable truth: Most AI underwriting is sophisticated pattern-matching, not real intelligence. Yes, many insurers now say they use AI, but only a small portion qualify as trailblazers. The rest are stuck in what I call AI theater—flashy demos that collapse under real-world complexity.
It should be no surprise, then, that a 2025 Capgemini report indicated that 88% of insurers recognize the critical future importance of advanced underwriting, yet only 17% have mature capabilities. When a model can’t explain itself in plain English, regulators get algorithmic word salad and underwriters go back to manual detective work.
Enter The Swarm: Collaborative Decision Intelligence
Think of the difference between hiring one “genius” who claims to know everything versus assembling a war room of specialists. In a swarm model, AI agents each focus on a different domain: financial, environmental, operational, behavioral.
They don’t just crunch numbers. They discuss and collaborate. They flag gaps in each other’s analysis. They challenge assumptions. And then, through structured consensus, they deliver a single risk profile.
Outside insurance, before the age of GenAI, swarm approaches have already delivered measurable gains. For example, swarm decision intelligence provided a 28-point improvement in financial predictions in a research study. In insurance, that same collaborative approach could mean underwriting that is more efficient, deeper and less biased.
Real-World Insurer Impact: Faster, Smarter, Fairer
Here’s an example of how swarm decision intelligence plays out. A 19-year-old with a gap in coverage, but a perfect driving record, would usually be flagged for manual review. With swarm decision intelligence, multiple agents weigh in at once—analyzing driver history, telematics data, neighborhood, technology available in the car and other key factors. Within seconds they reach consensus, complete with an audit trail that regulators can follow.
How Does This Tie Into Dynamic Insurance?
In my article on dynamic insurance, I laid out how risk pricing and policy terms are shifting in real time—driven by behavior, environment, usage and other evolving factors. Dynamic policies continuously adapt, rewarding safe driving, healthy lifestyles or low-risk behaviors as they happen. This shift will make insurance more responsive, fair and aligned with actual risk as it changes.
Swarm decision intelligence builds on top of that foundation. Swarm decision intelligence determines which data points matter most, how they’re weighted and why specific signals drive pricing or coverage changes. In short, dynamic insurance makes policies flexible, while swarm systems make them smarter. Combined, they pave the way for policies that not only adjust in real time, but do so with greater accuracy, transparency and trust.
What’s Holding Us Back
Multi-agent systems are heavy on computing, complex to orchestrate and tough to retrofit onto legacy platforms and fragmented data architecture. Regulators will demand clear audit trails, and underwriters may resist “black box” debates between machines.
However, there is nothing to fear, as swarm decision intelligence doesn’t sideline human expertise. It augments it. Underwriters become conductors of AI orchestras, not casualties of automation.
The Roadmap Forward
In addition to the steps I provided for insurers to start progressing toward a dynamic insurance reality, here is a rough blueprint for insurance carriers to get on the path of swarm decision intelligence.
• Zero To Six Months: Create a swarm decision intelligence task force, launch two to three pilot programs and audit current AI maturity.
• Six To 18 Months: Build infrastructure to clearly document, analyze and explain all swarm actions, train underwriters on orchestration and expand into more insurance lines of business.
• 18-Plus Months: Achieve full deployment, continuously refine agent specialization and establish market leadership.
Lead Or Be Left Behind
The insurance industry has always thrived on our collective wisdom. Those insurers who already have their data unified and are delivering enhanced wisdom have a head start over carriers with legacy systems as their engine.
Swarm decision intelligence gives our industry a chance to scale that wisdom beyond human operational limitations for the betterment of the insurance customer. In my view, the companies that embrace it now will define underwriting for the next decade, while the rest may be left wondering why their large expensive AI experiments still couldn’t match a small, well-equipped underwriting team powered by swarm decision intelligence.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
