AI Agents and Agentic AI: Demystifying the New Frontier

Marketfacts Home

Author

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

April 2025

The use of artificial intelligence (AI) within our industry is not new. We’ve been leveraging forms of AI, such as machine learning (ML) for automated and accelerated underwriting, for instance, for several years now. This new “Age of AI” was ushered in by the explosive growth of generative AI (GenAI), particularly by ChatGPT, in the first quarter of 2023. If 2023 and 2024 are remembered for the rise of GenAI, 2025 and onward will likely be remembered for the growth of “AI agents” and “agentic AI.”

The terms “AI agents” and “agentic AI,” despite being used synonymously, are related but not the same. Thrust into the spotlight by companies like Salesforce, NVIDIA, Microsoft, Meta, OpenAI, etc., AI agents and agentic AI are poised to become the next AI wave. These technologies can add significant value across the insurance value chain and within our industry ecosystem. Agents and agentic are new, and since our industry is in the learning curve with AI and GenAI, there is a significant gap in awareness, knowledge and understanding of this next AI wave.

Just like the rise of GenAI, this emerging concept has sparked both confusion and curiosity. In our industry, these terms are often misinterpreted as AI used to enable call center/customer service agents or insurance agents. However, AI agents can be used to augment, not replace, these human agents. Additionally, several factors have been fueling the confusion. With the explosion of AI, technology providers have sought to brand every software as a service (SaaS) product as “AI” or “AI enabled” over the past year. Now, this branding trend is swinging toward labeling every SaaS product or application programming interface (API) as an “AI agent.”

The overwhelming pace of AI growth has been exerting pressure on firms. As soon as companies develop a strategy to align their business objectives to one type of AI (e.g., ML, GenAI), a new advancement in the AI field emerges, causing firms to recalibrate their plans. Similarly, it is challenging to maintain the pace of educating employees fast enough to keep up with these advancements. Therefore, there continues to be a gap in the education and awareness of agents and agentic AI. This breakneck pace of change can also result in a phenomenon I call the “ostrich principle” — if I bury my head in the sand, things won’t change, and I won’t be hassled. However, firms will need to develop a foundational understanding of these AI advancements, as they will likely warrant attention and resources over the next couple of years.

AI Agents

An AI agent is an AI system that can perceive its environment, process information, make some autonomous decisions and take actions to achieve a goal. Agents can be rule-based — that is, following preset instructions — or more advanced, leveraging machine learning and reasoning.

An AI agent can be thought of as a digital employee — an individual contributor with a specific job to do — that can handle specific tasks independently, with little or no supervision. Common examples of AI agents include chatbots that handle customer service, recommendation engines that suggest products, or an AI system that monitors cybersecurity threats.

Agentic AI

Agentic AI takes AI agents a step further. Instead of merely following instructions, agentic AI can plan, make decisions and autonomously take action to achieve complex objectives, potentially across an entire ecosystem composed of several AI agents.

Agentic AI can be thought of as an AI manager — an orchestrator — that not only responds to its environment, but also establishes goals, makes adjustments based on feedback, and makes independent decisions in order to reach a defined outcome.

Agentic AI can be used to analyze your organization’s online sales strategy, identify gaps, run alpha/beta tests, and adjust the approach without any human intervention.

Illustrative Examples

Understanding AI agents and agentic AI through illustrative and relatable examples can help demystify them since these are still new concepts.

Illustration A:


AI Agent Think of an AI agent like your vehicle’s navigation system. It provides route guidance, but you still need to drive your vehicle to your desired destination.	Agentic AI Now, imagine a self-driving car. It knows your desired destination and can transport you there safely. Throughout the journey, the car makes real-time decisions without your intervention, such as finding the optimal route, avoiding traffic, conserving energy, and navigating around obstacles.

Illustration B:


AI Agent First, picture a stock market investment bot that you program to purchase a particular stock if the price drops below $X, or to sell the stock if it rises above $Y.	Agentic AI Now, envision a complex orchestrator that understands your stock portfolio goals, analyzes market trends in real time, conducts forecasting and model scenarios based on historical data (predictive analytics), and augments its prediction model by considering market fluctuations, geopolitical events, societal changes, and other factors affecting stock performance.

To further illustrate the impact and potential of AI agents and agentic AI, let’s explore the benefits and risks associated with these technologies, along with industry examples of their application in various areas. The following charts provide a detailed overview:

click below to open/close section

AI AGENT

AGENTIC AI

BENEFITS


Efficiency	Invaluable to automate repetitive tasks.	Invaluable for optimization of entire workflows.
Scalability	Can be very effective to scale and handle volume.	Expands and improves operations without extra human effort.
Decision Making	Excellent at analyzing data.	Proactively strategizes and executes based on insights.
Cost Savings	Significant savings through task automation.	Reduces human oversight in underwriting, claims, and sales.
Revenue Growth	Creates new automated sales and distribution channels.	AI-driven policy adjustments and sales optimizations increase conversions.
Risk Reduction	Automates quality control and quality assurance.	Prevents fraud and ensures compliance.
Customer Experience	Provides an omnichannel approach to engage customers how they choose.	Personalized, real-time policy management enhances engagement.

RISKS


Control & Oversight	Requires human supervision but follows rules predictably.	Can operate unpreditably, making independent decisions that may not align with human intent.
Bias & Fairness	Can inherit biases from training data but operates within constraints.	May reinforce biases over time as it self-optimizes, making oversight critical.
Security & Privacy	Risk of data leaks and cyberattacks but remains contained.	More vulnerable to adversarial exploitation and hacking due to its autonomy.
Adaptability Risks	Limited adaptability; struggles with new scenarios without retraining.	Can self-optimize incorrectly, leading to unintended harm.
Compliance & Ethics	Needs manual updates for regulatory changes.	May unknowingly violate compliance laws due to autonomous decision making.

EXAMPLES


Underwriting	Extracts medical records and scores risks.	Adjusts underwriting models based on new risk patterns.
Claims	Auto-fills forms and routes claims to adjusters.	Verifies, approves, and issues payments without human input.
Customer Service	Chatbots provide policy info and FAQs.	Proactively adjusts coverage based on life changes.
Sales and Distribution	Scores leads and suggests follow-ups.	Launches, tests, and optimizes entire sales campaigns.
Fraud and Compliance	Flags suspicious claims.	Investigates fraud and updates compliance policies dynamically.

In conclusion, the rapid evolution of AI technologies, particularly AI agents and agentic AI, presents both challenges and opportunities for our industry. As we navigate this new wave of AI, it is crucial to bridge the gap in awareness, knowledge and understanding. By embracing these advancements and integrating them thoughtfully into our operations, we can enhance efficiency, drive innovation and maintain a competitive edge. The journey ahead may be complex, but with a solid foundation and a proactive approach, we can harness the full potential of AI to transform our industry for the better.

0295-2025_MF_April_Cover Images_an_480x270.jpg