AI’s Potential to Shape the Insurance Industry: A Conversation with Cynozure

[Intro Music]

Suki Duper:
Good afternoon. My name is Suki Duper, and I’m the Head of AMER and APAC here at Tamr. Today, I have the pleasure of talking about AI and insurance with my colleague, James. James, would you like to introduce yourself?

‍

James Lupton:
Sure, Suki. I’m James, the CTO at Cynozure.

‍

Suki:
Great, James. As I introduced the subject, I think one of the key things our listeners will want to hear about is how insurers can leverage AI-driven insights, both today and in the future.

‍

James:
That’s a great topic, Suki. I’ve been spending a lot of time working with customers, not just in the insurance space but across various industries, helping them shape their strategies and deliver AI-driven solutions.

Before diving into specific use cases, I think it’s essential to define AI because the term means different things to different people. These days, most people immediately think of ChatGPT or large language models. While that’s an exciting part of AI, the scope is much broader. Some AI applications have been around for years and are already well-established in the insurance market.

For our discussion, I’ll be using AI in its broadest sense. I often use the analogy of a brain: the left brain represents classic AI use cases like fraud detection, recommendations, and predictions—deterministic tasks. The right brain represents generative AI, which deals with text, images, video, and audio—essentially creating, not just calculating.

‍

Suki:
That makes a lot of sense, James. It’s important to clarify that AI is different things to different people. At Tamr, we’ve been working in the AI space for over 12 years, primarily to help organizations unify and enrich their data.

AI can assist in today’s tasks, enhance efficiency, and offer new ways of solving problems. It’s important to highlight that AI’s value comes from its ability to support different use cases in various ways. As you mentioned, most people think of ChatGPT, but AI applications go far beyond that.

Speaking of applications, how do you see AI fitting into the insurer’s journey—from customer interactions to back-office processes?

‍

James:
AI is being applied across the entire insurance lifecycle. There are customer-facing applications, like call centers, and internal productivity tools, such as automating back-office processes.

Some applications, like risk modeling and pricing, have been around for years and use advanced analytical techniques, including machine learning. However, as we gain access to more diverse data sources—like third-party datasets, sensor data, satellite imagery, and weather data—we can move towards hyper-personalization.

For example, in health insurance, we could use data from an individual’s Apple Watch to understand their activity, stress levels, and sleep patterns, tailoring premiums and plans specifically for them.

‍

Suki:
That’s a great point. At Tamr, we’ve seen how critical third-party data is for insurers. Internal data alone isn’t enough—information from documents, contracts, and external sources is often locked away but essential for decision-making.

With the pace of change today—whether it’s customer phone numbers, locations, or behaviors—insurers must use external data to adapt their models. This is where AI can help. What trends are you seeing in other industries that might apply to insurance?

‍

James:
Across industries, I’ve noticed a shift from grand, transformative AI ideas to more focused, practical applications like improving individual productivity and governance.

Take generative AI tools like CoPilot. Many organizations are adopting these tools rapidly because the ROI is almost self-evident. For example, if a CoPilot license costs $300 annually, saving just a few hours of work per user justifies the expense.

However, I’ve also seen organizations fail when they attempt complex AI projects without addressing the fundamentals—like ensuring data quality and having the right datasets. This has led many to take a step back and focus on building a solid foundation before pursuing more ambitious AI applications.

‍

Suki:
That resonates with our experience. We always emphasize starting with high-quality data. If you don’t get your data right, you’ll struggle to achieve strong AI outcomes.

Do you see similar conversations happening with your clients about the importance of data quality?

‍

James:
Absolutely. Data quality is critical, but even more fundamental is ensuring you’re collecting the right data. With advancements in AI, data that was once considered irrelevant is now highly valuable.

For example, one insurance client partnered with a company that provides in-car sensors and cameras to monitor driving behavior. They offered a 30% discount to customers who used these tools, which helped the insurer gather better data for pricing and risk assessment.

This kind of data not only provides a competitive edge but also opens up new opportunities, like creating tailored products and re-engaging customers.

‍

Suki:
That’s a great example. Third-party data is becoming increasingly essential for delivering better customer experiences and driving innovation.

As you mentioned, AI models are improving rapidly, and insurers need to prepare by collecting and organizing data now to stay ahead. This brings me to another topic—trust in AI. Are you seeing concerns about AI trust and governance from your clients?

‍

James:
Trust is a significant issue, particularly for customer-facing AI applications like chatbots. Concerns range from regulatory compliance to potential reputational damage if the AI gives incorrect advice.

For example, I worked with a client who used a large language model to classify building descriptions for reinsurance purposes. Because this was an internal, non-customer-facing use case, they had the freedom to experiment and evaluate its effectiveness without fear of negative consequences.

On the other hand, customer-facing applications require much more caution, especially with regulations like the EU AI Act. Insurers must ensure their AI systems are explainable, well-governed, and used responsibly to maintain trust and avoid legal risks.

‍

Suki:
That’s a great point. At Tamr, we’ve seen how important it is to educate both customers and AI systems. AI is not a magic bullet—it needs good training data and continuous improvement to deliver the best results.

As we wrap up, what’s your vision for the future of AI in insurance?

‍

James:
I see increasing commoditization of AI tools. Insurers will need to decide what to build versus what to buy, as many AI models will be embedded directly into existing systems like CRM platforms.

The key differentiator will be the unique data insurers have. By focusing on collecting and organizing the right data now, insurers can position themselves to leverage these tools effectively and stay ahead of the competition.

‍

Suki:
I completely agree. The sooner insurers bring their data together and build comprehensive views, the better they’ll be able to deliver exceptional customer experiences.

James, thank you so much for this engaging conversation. For anyone who wants to reach out, you can find our details at the end of this presentation. Visit tamr.com for more information about our solutions, and cynozure.com for insights from James and his team.

Thank you again, James.

‍

James:
Thanks, Suki.

‍

[Outro Music]

‍