There has been much discussion in the insurance industry about artificial intelligence (AI) and how it can take on mundane, laborious tasks and complete them in a fraction of the time it would take a human.
But the technology is evolving rapidly. Agentic artificial intelligence is the next iteration of AI – effectively an autonomous, thinking robot that makes decisions, develops strategies, and continuously learns from the process. Expert predictions on AI’s future role in shaping insurance practices and the evolution of agentic AI were some of the hottest topics at the ITC Vegas insurtech event in October.
How Does it Work – and When Can it Be Used?
Agentic AI uses the same underlying technology as generative AI, but these newer models have “chain of thought” that allows them to think through steps to solve a problem. When a human asks the AI agent to carry out a task, it thinks about what steps are needed to accomplish that task. At that point, you can bring the human into the loop to consider whether those are the right steps, or you can let the AI proceed on the path it has determined and execute the solution.
To work effectively, it is essential that agentic AI is run on reliable data. It requires a repository of comprehensive, verified data to enable it to analyze a task and then execute an outcome such as creating a campaign that goes from marketing right through to sales. It can also assess the client’s needs and their budget, interrogate the available policies, and select the correct product. Agentic AI can even get in touch with the prospective client through voice agents that can realistically converse with humans.
Right now, there are agentic AI applications being developed for every part of the insurance ecosystem, from policy creation through to claims resolution. For example, there are AI producer agents available that target, launch, and optimize cross-channel marketing campaigns – with one click.
Human Touch
Typically, insurance producers need about 16 time-consuming manual steps to make a sale. These include identifying best leads, creating a marketing campaign, producing a sales brochure, social media activity and emails, all leading up to the meeting with the client and closing the sale. Agentic AI can automate the entire process that leads up to the meeting with the client.
While it can lay the groundwork for new business, eventually a human has to bind that business. US regulations state that only a licensed insurance professional can sell insurance and close the business. Agentic AI can take it right up to the point of sale when a licensed broker must become involved. There is no way, quite rightly, that a regulator would “license” a piece of software to deliver every part of the insurance cycle.
Overcoming Challenges
Agentic AI challenges are similar to those with traditional AI: ensuring it doesn’t hallucinate; making sure insurance producers meet regulatory requirements; data bias, and transparency. For these reasons, there will always be the need for humans to be involved to ensure decisions are made correctly.
As is the case with Gen AI, agentic AI cannot function effectively and safely if it’s basic fuel — data — is not clean, robust and reliable. It is also essential that this technology is tested on an ongoing basis to ensure that it is working correctly. There are advanced, upfront testing systems available as well as monitoring processes that will observe the agentic AI system and alert users if something is not working as intended. All of these checking tools require human oversight.
Data privacy is a major concern for some customers who are considering using agentic AI. To deliver the level of reassurance they require, it is important that suppliers have the right security and privacy guardrails to protect sensitive data about prospects or leads. Any reputable company has to protect data privacy, so this issue is not new. It is a little more technically complicated with agentic AI, but the regulatory requirements are the same.
It is also important to be transparent with customers about when and how you are using agentic AI. Often, regulators state: If you are going to use AI for underwriting, for example, you need to tell us how you got there to ensure the end result is not biased, false or unethical.
Productivity Tool
The widespread application of agentic AI is not a technological challenge – it is a change management challenge for insurtech companies to develop products that can give insurance professionals the confidence they need to accept it.
Rather than replacing jobs, agentic AI it will change them. It should be a viewed as a productivity tool – a helper that can take on the mundane tasks and allow people to be more efficient and get on with the more creative aspects of their work that require human insight and brain power.
There is a wide spectrum of sentiment in the market right now, ranging from those who adamantly oppose AI through to those who are extremely enthusiastic. This new, sophisticated technology has the potential to give companies the edge over their rivals. It is definitely coming, and in as little as one to two years, its use will be extremely prolific.
Ignoring AI is not an option, and those who are not currently thinking very seriously about how they will adopt this technology run the risk of being left behind by competitors.
Topics
InsurTech
Agencies
Data Driven
Artificial Intelligence
Market
Interested in Agencies?
Get automatic alerts for this topic.
