Smarter, Bolder and Faster: Agentic Power Unleashed

Smarter, Bolder and Faster:
Agentic Power Unleashed

MarketFacts Home

Authors

Kartik Sakthivel, Ph.D., MS-IT/MS-CS, MBA, PGC-IQ
Vice President & Chief Information Officer and Regional Chief Executive Officer – Asia West
LIMRA and LOMA

Nelson Lee
Founder
Infras (formerly iLife Technologies)

John Keddy
VP Insurance Strategy & Solutions, Lazarus AI

June 2025

As artificial intelligence (AI) continues to evolve at a rapid pace, the rise of AI agents and the emergence of agentic AI mark a pivotal shift that is set to transform our industry. These technologies represent the next evolution in how intelligent systems operate, collaborate, and create value across diverse applications within our organizations and the broader industry.

AI Agents

In the April 2025 edition of MarketFacts, we introduced the concepts of AI agents and agentic AI. An AI agent is a system that perceives its environment, processes information, makes autonomous decisions, and takes action to achieve specific goals.

AI agents, which range from chatbots to recommendation engines to cybersecurity monitors, operate like digital employees, often with minimal human oversight. They can follow preset rules or leverage both machine learning and reasoning to perform more advanced tasks.

Enter Agentic AI

Agentic AI takes this a step further, extending these capabilities by planning, making decisions, and taking autonomous actions to achieve complex objectives across the ecosystems of several AI agents. Agentic AI acts as an AI manager, setting goals, adapting to feedback, and autonomously reaching defined outcomes, such as analyzing online sales strategies, identifying gaps, running tests, and adjusting approaches without human intervention.

Traditional Versus Agentic AI

A simple way to distinguish traditional AI from agentic AI is by looking at their roles in decision making. For example, in traditional AI, underwriters can use an AI agent as a decision-support tool, directing it to flag high-risk applicants in the underwriting process.

In contrast, agentic AI, composed of multiple AI agents, can take on a more proactive and dynamic role. Agentic AI can independently prioritize applications, recommend pricing adjustments, generate follow-up tasks, and initiate communication with agents or advisors. Agentic AI can even prescribe a course of action for each applicant, such as when and how to communicate with them based on their preferences and preferred communication style (formal versus informal) based on applicant interaction patterns.

Agentic AI also unlocks a range of opportunities, including the ability to supercharge the automation of complex, cross-functional tasks, enhance operational efficiency, and shift both strategy and execution by anticipating market trends.

On the Flip Side

There are risks with agentic AI. Chief among these is the inherent unpredictability of agentic AI, which may autonomously take actions that, while technically correct, are misaligned with business intent or brand values.

Absent appropriate governance, agentic AI can make decisions with limited transparency, creating compliance and reputational risks. Additionally, when results diverge from intended outcomes, agentic AI can make it challenging to assign accountability and clear ownership within the AI ecosystem.

Now, as we approach mid-2025, interest in agentic AI is accelerating. But what does this technology really mean for the insurance and financial services industry? How can return on investment (ROI) be measured, and what does it take to be truly “agentic AI-ready”?

Measuring Agentic AI ROI

The challenging aspect of ROI measurement is finding a suitable use case where ROI can be effectively measured. Identifying applicable problem sets that would benefit from agentic AI automation is more than half the battle. Generally, the more tedious, manual, costly, and low risk a workflow is, the better suited it is for agentic AI automation.

Assuming we’ve found a great use case, the thought process to measure ROI should be straightforward:

How much faster does agentic AI complete this than pre-AI?
How much cheaper is the cost of AI ownership compared to the previous legacy workflow?
How does the accuracy of the new workflow compare to before?
Are there measurable improvements in revenue growth or user experience?
Does this let us scale our business growth faster than the traditional speeds of scaling talent?

Not all use cases must satisfy all four methods of ROI measurement. With AI agents, it’s important to focus on the problem and not the solution itself in order to maximize what you get out of the solution you choose.

Becoming Agentic AI-ready

Considering the frequent discussions around potential agentic AI uses, there has been little discussion around what makes an organization ready for these types of AI deployments. AI agents rely on data, access to different parts of the organization, and the logic behind an organization’s workflows to properly automate.

Central questions around AI-readiness are:

Is our data created centrally or is data created in many silos from many different sources and formats?
- If the latter, focus on unifying data creation and standardization in a reusable way.
Do we have a way to centrally orchestrate intersystem enterprise logic, so that agents can easily have automated rules they can follow both internally and externally without violating business processes and compliance rules?
Do we know the use cases that we want to automate and the intelligence we want to gather if we resolve our data fragmentation problems?

Once an organization has gone through a thorough inspection of the questions above and execution of its transformation toward AI-readiness, it is going to undoubtedly benefit much more from agentic AI adoption and enjoy much lower risk due to improvements in workflow orchestration infrastructure.

Upcoming Related Webinar:

Join Kartik Sakthivel, chief information officer, LIMRA and LOMA; Nelson Lee, founder, Infras (formerly iLife Technologies); and John Keddy, vice president of insurance strategy and solutions, Lazarus AI, as they discuss the differences between agent and agentic AI, emerging use cases in insurance, and practical insights from real-world examples.

AI That Acts: Discover the Future of Insurance

June 25, 2025 | 1–2 p.m. EDT