Agentic AI Exposed: Why Real-Time, Autonomous Intelligence Is Reshaping Insurance

The word “agent” used to mean a person selling a policy. Now, it means something far more powerful.

In this episode of Insurance Unplugged, Lisa Wardlaw sat down with Kyle Geoghan , CEO and Co-Founder of Indemn , to unpack what Agentic AI really is, why it’s misunderstood, and how it’s beginning to reshape the insurance infrastructure from the inside out.

Forget everything you think you know about AI. This isn’t about chatbots or simple workflow automation. It’s about autonomous digital agents that operate with intent, learn from your best people, and function in real-time—with or without traditional APIs.

Let’s break down the fundamental transformation happening right now.

What Is Agentic AI—And Why Should Insurance Care?

“Agentic AI doesn’t just complete a task—it interprets intent, maps a strategy, and adapts in motion.” - Kyle Geoghan

Agentic AI is more than just another automation tool—it’s a fundamental shift in how insurance companies operate. Instead of relying on humans to process information, analyze data, and execute tasks, AI agents can now do this autonomously, with minimal oversight.

Geoghan defines Agentic AI in two tiers:

• Shallow Agentic AI: Executes predefined tasks using curated knowledge. It can retrieve information and automate structured workflows but doesn’t make autonomous decisions.
• Deep Agentic AI: Accepts a goal and autonomously works to accomplish it. This is where fundamental transformation happens—AI agents that don’t just follow rules but make decisions, adjust strategies, and learn from experience.

“Unlike RPA, Agentic AI operates with autonomy—it doesn’t just execute tasks, it makes decisions.” - Kyle Geoghan

This is not the same as robotic process automation (RPA). AI can react dynamically to new information, making it far more effective than static, rule-based automation.

This is a game-changer for insurance. Instead of requiring manual intervention to handle policy adjustments, claims processing, or underwriting decisions, AI agents can act in real-time, ensuring that insurance businesses move faster, operate more accurately, and deliver a frictionless customer experience.

Why APIs Aren’t Enough Anymore

For years, APIs have been the backbone of digital transformation in insurance. But here’s the problem: APIs weren’t built for real-time, autonomous decision-making.

APIs function on a request-response model—you ask for something, and the system returns a pre-defined answer. That’s fine for structured transactions, but it falls apart when you need real-time adaptability, multi-step decision-making, or cross-platform intelligence.

“APIs were built for form submissions, not for autonomous decision-making.” - Kyle Geoghan

In contrast, Agentic AI can operate across multiple systems, interpret intent, and execute actions dynamically. It doesn’t wait for human-triggered API calls; it proactively retrieves information, processes decisions, and interacts with other AI agents.

For insurance companies, this means:

• Faster, real-time policy adjustments without manual intervention.
• AI-driven claims processing that doesn’t require back-and-forth approvals.
• Seamless integration across multiple systems without rigid API dependencies.

Rethinking the Data Problem: AI Agents Don’t Need a Common Format

“AI agents don’t need perfect data—they work with what’s available and adapt in real-time.” - Kyle Geoghan

One of the biggest myths in insurance is that data standardization is the key to AI adoption. The industry has been chasing the dream of a universal file format for years, allowing every system, carrier, and broker to exchange data seamlessly. But here’s the reality: that future isn’t coming anytime soon, and AI doesn’t need it to function.

Rigid data structures don’t constrain agentic AI. Unlike traditional automation tools, which require carefully formatted inputs, AI agents can navigate carrier portals, extract relevant information, normalize data from multiple sources, and surface only the fields that matter—without requiring a predefined, industry-wide standard.

“If you’re waiting for the industry to standardize data formats, you’re already behind.” - Kyle Geoghan

Instead of spending years and millions of dollars trying to clean up every dataset, companies should ask a different question: “How can AI work with the data we already have?” The answer lies in AI’s ability to interpret, adapt, and make real-time decisions rather than waiting for the perfect input.

The Real Danger: Automating a Bad Process Isn’t Innovation

One of the biggest mistakes companies make when adopting AI is automating broken processes instead of rethinking the entire workflow.

“Don’t repeat the RPA mistake. Start from first principles, not from what’s already broken.” - Kyle Geoghan

For example, if your underwriting process is slow, inefficient, and reliant on manual intervention, simply adding automation won’t fix the underlying issue. Instead, AI should be used to redesign the process from the ground up, eliminating inefficiencies rather than accelerating them.

True innovation requires examining the entire operational model—not just digitizing old workflows but replacing them with AI-driven systems that think, act, and optimize autonomously.

“Insurance doesn’t need more data lakes. It needs AI that can extract meaning from what’s already there.” - Kyle Geoghan

Start, Stop, and Continue

Start: leveraging AI that can interpret unstructured data in real-time. Instead of waiting for perfect data formats, companies should implement AI that can extract meaning from emails, PDFs, and legacy systems without requiring human intervention at every step.

“Stop automating broken processes. Start rethinking them from the ground up.” - Kyle Geoghan

Stop: assuming that every data source needs to be perfectly structured before AI can deliver value. The belief that AI is only effective when data is harmonized is outdated. The reality is that Agentic AI works with what’s available, adapting to inconsistencies rather than requiring rigid inputs.

Continue: investing in AI-driven workflows that improve over time. AI systems learning from real-world data inputs will consistently outperform those requiring pre-formatted structures. Instead of waiting for industry-wide alignment, deploy AI now and let it refine itself as it goes.

“The goal isn’t to replace people—it’s to free them from repetitive work so they can focus on what matters.” - Kyle Geoghan

The companies that embrace this shift will move faster, reduce operational costs, and unlock AI’s full potential—today, not years from now. Those who continue to wait for a standardized ecosystem will be left behind.

Final Word: It’s Not Hype—It’s a New Operating Model

Agentic AI isn’t just another tech buzzword or incremental improvement. It’s a fundamental shift in how insurance operates.

• Replacing latency with real-time decisions
• Replacing static portals with dynamic systems
• Replacing repetitive work with trained AI agents that think and act

This isn’t about adding AI to existing processes—it’s about redefining how insurance works from the inside out.

And the companies that embrace this shift today? They’ll be the ones leading the industry tomorrow.

For more, click here to listen to the full episode of "In the Hot Seat" from Insurance Unplugged with guest Kyle Geoghan and host Lisa Wardlaw. Thank you to IRYS Insurtech for sponsoring this season of Insurance Unplugged.

Subscribe to Insurance Amplified on LinkedIn and listen to Insurance Unplugged to stay tuned for future episodes and editions!