⚠️ Unsupported Browser

Your browser is not supported.

The latest version of Safari, Chrome, Firefox, Internet Explorer or Microsoft Edge is required to use this website.

Click the button below to update and we look forward to seeing you soon.

Update now

The Future of AI Agents in Insurance Customer Service

Kylie Whitehead
18 Aug 2019 - 8 minutes read

We’ve recently spoken with several big name insurance companies about how they can begin to implement AI in their customer service departments.

In this post, we’re going to share some of the use-cases we’ve been thinking about, and how insurance companies might go about implementing them.

Identification and Verification

Existing customers will always be required to go through ID&V checks before accessing or editing their information. As security regulations continue to get tighter, there will be more restrictions on how contact centers need to deal with personal information.

Using conversational AI to identify and verify your customers completely removes the need for your customers to divulge their personal details to your agents. What’s more, the AI is able to check postcodes and addresses against national and international databases, and as it can integrate with your CRM, it’s not necessary for any of the information given over the phone or online to be written down or stored in order for it to be checked.

AI-powered ID&V can handle common mishearings and various accents effortlessly, preventing customers from needing to repeat themselves. If you’re interested in using conversational AI for ID&V, check out our recent blog post on just that Using Conversational AI to Manage ID&V in Contact Centers.

Policy changes and updates

Any industry that relies on large amounts of customer facing documentation will have experienced the difficulty of re-training customer support agents on this new information.

Good AI agents will be far easier to train. At PolyAI, we use a process we call content programming to drop new information directly into the model that powers the AI agent. Because the model has been trained on millions of conversations already, it’s able to understand how this new information fits in, and can start using it in conversation immediately. You can learn more about how AI agents respond to customer queries in our recent blog post.

Frequently asked questions

Using intent detection, AI agents can understand complex queries and draw out what it is customers are actually asking for.

Our conversational search engine makes our AI agents perfect for handling frequently asked questions. Because the agents have your knowledge base at their fingertips, they can answer any customer queries that are covered in the knowledge base, regardless of whether or not you’ve specifically trained them for that specific query.

Making claims

When customers make insurance claims, they have to provide a substantial amount of information regarding their circumstances. Agents must get the correct information from customers and add it into the correct database. This process is very much like filling a form, but using voice. Conversational AI is incredibly well equipped to handle this type of conversation.

When a customer phones up to make a claim, they’re likely to give a lot of information up front. For example, somebody might say, ‘my iPhone X was stolen about half an hour ago at Kings Cross train station.’

In this example, the customer has explained:

  • A theft
  • An iPhone X
  • About half an hour ago
  • Kings Cross train station

For a human agent following a script, it’s difficult to record all of this information immediately, and impossible if the software you’re using requires information to entered in a specific order and across multiple screens.

AI agents, on the other hand, can pick out all of these relevant pieces of information and drop them in the correct slot off the bat. (Check out our previous post explaining how this is done). This way, the AI can fill in the gaps (e.g. ‘what colour was your iPhone X?’, ‘did you report the theft to the police?’) and the customer doesn’t need to repeat the information they’ve already provided.

Unlike pre-scripted chatbots, AI agents can handle non-linear conversations, where customers can ask questions along the way (e.g. where do I find my policy number?), change their mind (e.g., “oh sorry, it’s with my married name, Mrs Jones”) or correct the agent (e.g., “no it’s J O N E S”).

Taking payments

Recent changes to the Payment Services Directive (PSD2), are making it easier for product companies to gain access to customers’ bank accounts (with their permission of course) in order to provide new financial services. While this is great news for opening up an industry that has historically been in the hands of a small number of enormous institutions, it means any company handling payment information will need to jump through more security hoops than ever before.

Conversational Al is by far one of the most secure ways to take personal information. Imagine that a customer paying a premium, or upgrading their account, does not have to give their bank details to a human at all. They’re simply handed off to an AI agent, who takes the details, passes them over to the payments provider, and forgets them. No one sees the payment details, and they’re not stored anywhere.

Unlimited capacity, in any language, on any channel, at any time

Unlike your human agents, AI agents aren’t limited in their capacity. They can deal with multiple customers simultaneously, without compromising efficacy. Once you have the core technology in place, you can deploy conversational AI across any conversational channel, from phone to webchat to smart speaker to text message; so you can speak to your customers using the channels they prefer.

And because conversational AI handles speech as a mathematical process, it’s much easier to deploy versions of your AI agents in multiple languages than it is to hire teams of native speakers around the world.

Knowing when humans are best

Not all interactions are best suited to AI. While AI agents should always be used to deal only with appropriate intents, conversations can still end end up going off on tangents. AI agents should be trained to know the limits of their own capabilities, and can hand off to a human agent at any point such as if the customer asks to speak to a human, if the agent fails a certain number of times consecutively, or if the customer uses certain trigger words.

The future of AI in your contact center

While fully autonomous assistants are still some way off, conversational AI can be an effective addition to your customer service team. AI agents are perfectly positioned to handle repetitive tasks like form filling, ID&V and taking payment details.

If you’d like to learn more about implementing conversational AI in your customer service department, get in touch with PolyAI today.

Get more like this!

Get the latest on conversational AI for customer service

  • This field is for validation purposes and should be left unchanged.