Financial services giant Prudential crafts agentic AI strategy

Prudential Financial might have been founded in 1875, but its IT stack is definitely next-generation.

The 150-year-old company has 38,000 employees who serve 50 million customers in more than 50 countries. Bob Bastian, chief information and technology officer, leads the company's agentic AI initiative, which is built on Salesforce's Agentforce for Financial Services.

We discussed with Bastian the company's deliberate approach to rolling out AI services for its various insurance, retirement products and mutual funds.

How would you describe Prudential's AI strategy?

Bob Bastian

Bob Bastian: Because we're a regulated business, we give a lot of thought to how we're using AI. Our main strategy is thinking about how to build governance so it does exactly what we need it to do for our customers and stakeholders, but at the same time, it's protected, it's governed and we think about data management right out of the gate.

Then we say, "Where would it provide the most value for our customers or advisors?" We've created various use cases that help us understand that value proposition. Or, in the case of something like underwriting -- something we've been doing for quite a while -- we have multiple tests [of AI tools] against our underwriting to make sure that we're getting valid results.

When you saw generative AI for the first time, did you immediately see use cases?

Bastian: I enjoy technology overall. I just love to understand how things work. My first thought was, "Okay, I need to understand more about how this is coming together."

After understanding about LLMs [large language models] and RAG [retrieval augmented generation] and all the other technical details that go into it, I started to think about how this can apply itself to the insurance, annuity and advisor spaces. Everything that we've come up with focuses on a human in the loop, just because of how GenAI is structured. It's important for us to make sure that we still have humans thinking about what is happening. We are a company that generates a lot of documentation. We want to ensure that we get the right policy, contract and type of products out there in production. It necessitates a lot of information, and GenAI can help pull that information together. Many of our use cases are set up around that.

What are examples of operational use cases?

Bastian: We have a use case for group insurance claims. They're different based on each company's contract and the company's state regulations. So if I'm on the phone with a customer, I have to go out to different places to pull that information together. With what we've done with GenAI, we know who the customer is, we know what company they work for, and we know what state they're coming from. We gather all that information so that, instead of trying to find different documents, CSRs [customer service representatives] can focus on the customer and be very empathetic.

We also have another use case for advisors. It pulls together information about who they're meeting. We have wholesalers [who support financial advisors] as well as advisors. In the wholesaler space, we're pulling together information about the people they're meeting with so that they can come well prepared. That makes them a lot more productive.

Do you employ multiple LLMs for different functions, or are you a one-LLM shop?

We give a lot of thought to what the right LLM is for the job.

Bastian: We are not a one-LLM shop. Different LLMs are optimized for different things. We give a lot of thought to what the right LLM is for the job. We also give thought to how we're going to apply it; what is the right risk for our customers, stakeholders, advisors, whomever else. So we're very protective how we're using LLMs, which ones we're using and which ones are the best for any given use case.

What are you doing with Salesforce agents and Financial Services Cloud?

Bastian: They have done some great things in our wholesaler space, as mentioned above. We consolidated our customer service across all our businesses onto Salesforce. We consolidated all our claims -- except for disability claims -- onto Salesforce. Our retirement business, our U.S. life insurance business, group insurance business and everything is on Salesforce.

On the sales side, we're using agentic AI to help with call logging. If a CSR meets with a customer or a wholesaler meets with an advisor, it logs what happened during that meeting. The agents can go off and deal with follow-ups. Agentic is great at summarizing how we met with [an advisor], how they want to do these types of cases in the future and what follow-ups are needed, then it will set those up in Salesforce and deal with those background pieces. We think we can save at least half a day per week, just on different follow-ups and setting up other things in Salesforce. So that's a great use case beta that we're starting.

On the service side, there's a lot that goes on after a call. When a customer calls up and has a service request, there are a bunch of other things, different follow-ups, different background things that happen with administration systems and others. We're in a beta there where an agent can do some of those things. [This allows] the CSR to get on the next call and be more empathetic to the conversation that they're having -- as opposed to having to think about the call log, follow-ups or the things that they have to do in the background. This agentic force helps us get through many of the background [tasks] so that our advisors or CSRs can stay focused on the customer.

What advice do you have for your peers experimenting with or evaluating agentic AI?

Bastian: One thing is to keep customers at the heart. How do you make your CSRs better with the customer? How do you make them more empathetic? How do you make it a great relationship, so when a client calls up Prudential, they feel like the agent is listening to them? Design your agents keeping in mind how to enable better interactions.

The other thing, especially for our industry -- whether it be bank, financial services or asset management -- is that we're regulated, so you've got to think that through first. Don't go off and create a bunch of agents that do a bunch of things that could get you into trouble. Think through the governance before you get down the path of thinking about agentic.

If you focus on a customer at the heart of it and you really balance yourself with governance and regulatory, that space in the middle that allows the human to do the best that they can when they're on the call or when they're meeting with an advisor. That's the path to greatness on some of this.

Don Fluckinger is a senior news writer for Informa TechTarget. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.