A Hurricane Hits Boca Raton

Embark on a journey into the future of claims control with our session, 'The Claims Control Center.' Step into an AI-driven Disaster Response Command Center, where cutting-edge technologies like drone footage, 3-D imagery, IoT sensors, and real-time data feeds converge to revolutionize disaster response and claims management. Follow the realistic hurricane scenario as it descends onto Boca Raton and delve into round table discussions centered on disaster response, claims evaluation and management strategies. Explore how technology is reshaping the insurance landscape for proactive and streamlined claims handling.


Transcription:

Gina Hardy (00:10):
Well, thanks everybody for being here. We really appreciate you. A hurricane hitting Boca Raton is a little less scary for us than a hurricane hitting North Carolina. So let me tell you who we are and then we are going to lead you in a dialogue. So I hope all of you brought a lot of ideas and innovation and having this is more of a discussion as opposed to we talking to you. So we'll talk through some of the ideas and we'll also talk through some of the risks that some of the new technology could create for carriers as we work through it.

(00:48):
I'll introduce each one of us. My name's Gina Hardy. I'm the CEO of the Residual Market in North Carolina. I have with me today Tekeela Barnes, my CIO. I also have Vicki Fisher, Director of Claims Operations. I have Christine Wertis, our Claims Manager, and Robert Switzgable today is working with us. This is a look at our insured values and North Carolina, and so you can see the dark areas or areas where we insure anywhere between 10billion and 29 billion. I wanted to show this slide, so to talk about the fact that we live this scenario every day in planning for this scenario every day. What happens if we have a cap five hurricane that was to impact the coast of North Carolina and how can we restore lives as quickly and as efficiently as possible?

(01:47):
We did have Hurricane Florence in 2018 and Hurricane Florence had a hundred thousand claims between our two associations. We were able to settle those claims within 90% of the claims within 60 days and with less than 50 litigated files. So for us, being able to quickly move to have claims settled is extremely important. These are some of the top hurricanes that we have seen impact the us and you can see that most of these hurricanes have occurred since the year 2000. And this is our outlook this year and I think for all carriers, this outlook should make us nervous Going into this year, it's projected to be 85% above normal hurricane season, and we've already seen our first name storm. They're projecting 17 to 25 name storms, 8 to 13 hurricanes and four to seven major hurricanes. When we got the assignment to help lead the discussion today, these were the questions that we were going to tackle is one, AI and real-time data. How do we use it to prepare for a hurricane? How do we use it for FNOLs and assignment of claims? What real time data we can use for data assessments? And then what are you seeing in future assessments of claims? So with that, Vic, I want to let you and Christine start.

Vicki Fisher (03:31):
Hi everybody. Welcome. I appreciate you being here. We really want this to be a round table discussion where we get all of your ideas and maybe you guys share some of the things that you've been doing to help everyone come up with some good ideas. So we'll start with the first question about AI and real-time data to help prepare. I'll kick off a couple ideas and then I'd like to open it up to the room to see if any of you have things that you've actually implemented. One of the things we've thought about is obviously getting a lot of data about our market and where the storm's coming in and what policy holders may be hit by a storm and then want to have kind of open discussion with you. If you get that data, what do you do with it? And is anyone doing anything with that now? Nobody's using it yet. Okay, so that's Go ahead.

Audience Member Erica K (04:20):
I was only going to know that.

(04:23):
Sorry. We're using just based off of the weather, the weather data, we are bouncing that against a database and just notifying policy holders that there's an incoming threat and also how to prepare for the storm and putting out social media alerts and things of that nature if we know where it's coming and who's potentially impacted.

Gina Hardy (04:49):
Is everybody able to hear

Audience Member Erica K (04:51):
That?

Gina Hardy (04:53):
Oh, you cannot hear. Do you mind sharing real close to the real close? You can stand up.

Audience Member Erica K (04:59):
Hi, Erica Kidder. Yes, so I was noting that very simplistically we are using data from the Weather Alert system and able to notify our policy holders that there is an incoming threat and email them and contact them so that we are able to give them some preparation. What do I want to say? Just help prepare them and help give them some tips on how they can be better prepared for the storm. Not perfect, but continuing to work through that and hopefully grow that in the future.

Gina Hardy (05:49):
Anything to start?

Vicki Fisher (05:58):
We are not.

(05:58):
Is there any companies that are doing any type of flyovers to get data damage, to get damage information on what the properties look like before the losses versus after the storms? Has any companies started to do that?

Audience Member Dana (06:14):
Yeah. Hi, I am Dana from State Farm and so we use external and internal data to determine how many homes and lives we insure in the particular area. And that helps us know what capacity we need to deploy in our CAT teams. So really staffing and resourcing and how many bottles of water and care bears to take with the teams, all the resources. That's one application. And then you just mentioned something that left me starting the client talk. Oh yeah, yeah. We're not doing that yet, but it is definitely something that we're exploring for sure.

Vicki Fisher (06:59):
And are you doing anything with looking at pre-loss condition versus after loss?

Audience Member Dana (07:04):
Oh yes, that was the other piece, yes. We're buying, we partner with an early stage company that provides that data. So we have aerial data, drones and satellites and flights. They capture that data and share it with us. Yeah.

Vicki Fisher (07:26):
Do you know if you've found after a storm when you're looking at your damages after the fact that that pre-loss dentist has been helpful?

Audience Member Dana (07:34):
Well, yeah. I mean we fly our own drones after the loss has happened as well. We have a whole team of pilots that are able to fly drones over the neighborhoods that are damaged or entire geographic areas. So yeah, we use it a lot.

Vicki Fisher (07:51):
I do have a question. North Carolina has kind of strict laws on using drones and getting photographs of the next door neighbor's house versus yours. How have you guys handled that?

Audience Member Dana (08:01):
I'll have to ask my claims team, but we can connect you. Don't me to put you on the spot. No, I'd be happy. We were one of the first insurers to get the FA licenses to fly beyond line of sight and over people and property. So we've been doing it for quite some time, probably 10 years now at least, maybe longer. So I'm definitely happy to put you in touch with the right folks if you want to learn more about what we do.

Vicki Fisher (08:24):
I would love that. Has anybody else dealt with that where they have to get permission and get the right authority to be able to fly over and get the drone footage? Because in North Carolina they're very strict. You can't get your next door neighbor's house in your drone footage. Even if you go out to look at that one house, you have to make sure the drone footage is done in a way where your next door neighbor's house isn't in any of the footage. Has anybody else dealt with that? Something we're going to keep working on?

Christine Wertis (08:53):
What about post-event? So you mentioned a little bit about utilizing drones to determine here's where our damaged areas are. Then using that to say, okay, well let's deploy a group of adjusters or maybe customer service to help those insureds who need that help back there.

Audience Member 3 (09:20):
I still see Harris. So we do, we have pre photos of properties and then we fly over and we see we the greatest damages and we use that to get our adjusters to the most heavily impacted areas and then move out from there. It helps us really get the situation under control and taking care of those most in need. So we do that and then we do compare the before and after because you could have a 30 5-year-old roof that you're dealing with. So that's what we do and we don't really have that, not that I know of that problem of not being able to look at a neighbor's property because lori's here from personal lines, we know that some of the risks to the property is actually what's happening in their neighbor's yard and how well they're taking care of their trees and what have you. So we do look at that. I think,

Audience Member Mike Pigeon (10:21):
Hi, I'm Mike Pigeon with Duck Creek. We have a partner we work with and we work exclusively with them during Hurricane Ian as it came in. And one of the things they were able to do was pre-storm as the cone was here and then it got smaller accurately gave us what was happening, storm forecasted, storm surge and windstorm. But we were able to put stage adjusters outside of the cone and then as the storm came through, we were able to look at the damage and then move those adjusters in from a top down level Eagle Eye view. Also, we were able to use whatever was available, which was at the time Noah, so we could go down a certain amount of feet into a neighborhood. But then we were able to enter Noah information, which gave us even further down. So the cool part about that was if it existed out there in a third party, we could just add that to this platform and get all the way down into a rooftop type of situation.

(11:25):
And then we actually had pre and post when you could go back into another database and get the pre what it looked like pre and post. And the clients that were using this at the time were able to also look at all their piff along the coastline and accumulated value of that piff. And then so from a standpoint of reserving, that was helping them too. So it helped in claims, it helped in the underwriting and reserving aspect of it, but the products out there are getting much, much better with that. So we just did it with one particular with partner that has partnered into our network, so I don't want to push a product, but that was really good.

Gina Hardy (12:04):
Did they end up helping them with factor reserves too? Did they use this product to put up their factor reserves?

Audience Member Mike Pigeon (12:12):
It was all the data from the actual insurance company. So we uploaded that and that was the cool thing about it could upload it in three seconds. So whatever data you can upload into the system, you can manipulate after the fact. So it's really, if you can upload it, you can calculate it and spin it out. That was the beauty of it. It's very simple. It wasn't limited into what you can upload into it.

Christine Wertis (12:41):
So what about that FNOL intake? What have you guys done to automate or utilize AI for FNOL intake, if anything?

Vicki Fisher (12:57):
Are anybody using chatbots or anything to do FNOL or are you doing 'em over the phone or online? What are you guys using?

Christine Wertis (13:07):
Do you rely on your agents? Do you have independent agents direct? Right.

Audience Member 5 (13:20):
It's still new in the industry as you know. And the automation for FNOL, you can use many, many tools, but I think it's challenging for the company to adopt the new technology because the data is not always in the right place. And there are a lot of challenges in terms of cleaning up your data, making sure it's safe as well as choosing the right partners and upgrading the workforce to get there. So the question I have for you all is what do you think you want to do first as a use case for those of us that haven't implemented anything yet?

Vicki Fisher (13:55):
I could take a little bit of that. So one of the first things we developed was just a simple online report of loss where the insured can go to our website and report the loss, but we'd like to work towards develop it more of an IVR or phone one where the system will drive that report of loss because after Hurricane Florence, we had so many phone calls that if we could get some of those insureds to report those online or through even on a phone IVR, it would save us so much time. And so I think an IVR type of driven and we haven't developed that yet, but we do have the online and so we're working towards that.

Tekeela Barnes (14:34):
Yes, in our new phone system we implemented about a year ago, we're still looking into that capability because as Vic said, we had thousands of calls a day during Hurricane Florence in 2018 and yes, yes.

Audience Member Erica K (14:58):
I'm so sorry, Tekeela, I did not mean to No worries. To cut you off. This is for you at all.

Tekeela Barnes (15:04):
Exactly. This is for you.

Audience Member Erica K (15:08):
I had a question about FNOL and your automated claim process that everybody obviously pushing folks to notify the carrier first notice of loss online is obviously more effective. How do you triage based off of the severity, right? So somebody might be very comfortable with entering in their FNOL if there was a minor collision or some minor property damage or something to that nature. But how do you balance when you have to have that empathetic touch to that consumer at that most critical time? When we're constantly driving AI and we're constantly driving the technology, how are you making sure that those AI, the chat bot, all of the digital bells and whistles, how do we make sure that that's actually not being incredibly disingenuous to somebody who just had a very significant loss or a tragic loss of life?

Gina Hardy (16:17):
That's a really good question that she put out to the group. What do you guys think?

Audience Member 5 (16:25):
This is where AI can really shine because when you're getting thousands of calls a day and your agent are working 24 hours to process these claims, and the way you can go about it is that if you can implement gen AI that would understand sentiment and really do the real time analysis and cue the human agent on, ooh, here's a problem, what do we do?

Tekeela Barnes (16:51):
I love that.

(16:52):
Thank you. And that is so true because right now our current phone system just in day-to-Day phone calls, it has the ability to understand tone, whether it is sadness or I rightness or disrespectful, and it can show scripts to help our agent manage that. And it can also cue the supervisor that they may need to jump in.

Vicki Fisher (17:20):
But I really like your suggestion because yes, everyone wants to do quick and easy and the younger generation wants everything online, but sometimes they need a human and they need someone to help them through If their house is unlivable, they want someone to talk them through what to do next. And so I'd love the fact that it could read that sentiment and get them to a live person and then if it's a small claim and they don't need a live person, they just move on. Yeah, I love that. How about for damage assessments and just resolution of the claim? So there's obviously a lot of steps in resolving a claim, and so what do you guys obviously think some of you're using drones, we've heard that a couple of groups. Are you having good success with drones? Yeah, kind of.

Audience Member Dana (18:12):
Honestly, the challenge has been a human challenge, which is just trusting the data so we can get a wonderful drone picture that tells us what hail damage was and exact measurements and everything. But there's still this natural tendency to want to send a human up a ladder and go validate what the picture tells us. So that's just been an evolution over time of just trusting the data. I was also going to add on the FNOL discussion. I mean we still have 19,000 agents and thousands of adjusters who go to the sites of major events like we're discussing here. So that might be the human, the walk-in or drive-in part of the gathering the data. But we also have built into our mobile app the ability to upload pictures even of an internal damage like a kitchen fire or something like that. And you can have a conversation with the adjuster about the video that you took of your kitchen just to assess the damage for an internal, but that's of course not a major event like a hurricane where your home is no longer on its foundation. That's kind of a different conversation. So

Gina Hardy (19:25):
What about process oriented as you're looking at settling claims, what about some process oriented things that AI might be able to help you with? So think about some ways that you're currently using or some ways you're considering using it. Any commentary on that please?

Audience Member 3 (19:48):
I mean it's the obvious stuff like the document ingestion is a big part. There's also the part your retirement empathy and human interaction, allowing your rep to be able to be focused on the person on the phone a hundred percent and really hearing what they're saying and what's going on. It's about capturing the notes. That's such a low value kind of, but huge for customer interaction kind of tasks. So if you can take all that note taking off the responsibility of the person, the rep on the phone and use AI to take that capture it, it really could make a difference on what that interaction's like and how long it takes to get to what they really, we were always wondering whether people were missing things as they're trying type at the same time they're talking to somebody. So I think that that's a huge opportunity.

Gina Hardy (20:49):
I think so too. And I also think what you mentioned at the beginning, which is document classification, so that the files are such that when you have a large storm, you normally start out with a very large number of adjusters and then you start pairing those adjusters down and having the most organized file possible so that the documents are read, they're appropriately classified and organized so that when on a CAC claim you could potentially have several adjusters as the claim continues to process through. And I think that was a very valid point. The one thing that we've been pursuing. Do you want to talk a little bit about letters and how we can do letters more effectively so that we can better explain if something is covered or is not covered and how we could cover them?

Vicki Fisher (21:43):
So for us, because strictly property, we obviously quote the policy a lot when we're writing our letters as far as if it's a partial denial or denial, and those can get really cumbersome for the adjuster. So we've been looking at more and more ways to automate that. And we've talked a couple of vendors about doing more with AI where they can type a short little thing and it'll pull up the language that they need out of the policy forms and stuff because now they have to go find it and cut and paste it in and it's very time consuming. Is there anyone in the group who's done anything more robust on that? I know a lot of companies use templates that start the process for the adjusters, but sometimes there's so many different scenarios that you have to pull the policy language for that a template just can't cover every scenario. So that's one of our challenges that we're trying to look for. And I've been talking to vendors here that we're trying to solve for

Gina Hardy (22:34):
How can you have a co-pilot on it? Go ahead, Sicily.

Audience Member 3 (22:39):
The one problem around that that we've been grappling with is which policy term are you covering and being able to understand the timing of the claim, the timing, the language and all of that. I'd be interested to hear if anybody's,

Gina Hardy (22:57):
Especially as policy forms evolve and change and modify. Exactly. And especially if you have a claim that's held out for several years that's not getting settled.

Audience Member 3 (23:07):
That's exactly it.

Gina Hardy (23:07):
Those are huge considerations to make sure that you get that right.

Vicki Fisher (23:11):
Yeah, I think we'd have to read a lot of data in your claim and policy file to make sure that it's grabbing the right forms, the right version of the form and the right limits and all those things. So I think it's going to be a little bit of a big build, but I think it would have a big bang as well because if you're writing a complicated denial, it can take you 45 minutes to an hour and the adjusters are pulling all this language and it's really time consuming. So if any of you come up with great ideas, find me. I want to know them.

Gina Hardy (23:41):
That's good. That's good. What about any other process oriented? We talked about documents, we talked about note taking, what about other process oriented as we go through this? That's a really good one. Signature verification. That's a really good one.

Audience Member Dana (24:00):
And then we leverage ai. We built a proprietary voice bot that enables our auto customers who are using a rental car for a period of time to call in and just extend their rental. So it's a really simple little thing, but was consuming a ton of time in our contact centers. And so we honed in on that problem. What we learned 10 years ago when we started down this path of building our own bots, we built our own because we found that the libraries, the language models that were behind these bots didn't really speak insurance language and they had a really hard time with alpha alphanumeric characters. So I know these problems have been solved of late, but we've built our own to solve that specific problem. So we have a strategic alliance with Hertz, and so we leverage that to extend their hertz rental and it saves our contact centers time and it's been a really good customer experience as well.

Gina Hardy (25:03):
One thing, go ahead.

Tekeela Barnes (25:04):
I would like to speak to the customer experience because one of your insurers hit me and I actually utilized the bot and it was a great experience for No, it really was a great experience for me. That's why I wanted to share it here. If it were not, I would not have put you on the spot, but it really was so I can attest that it's working.

Gina Hardy (25:30):
I think one thing that we look at too is on the property side, we have similar vendors that we need to have that relationship with and being able to assign those and that certainly helps. How about other things, Vicki, as you go through that brings the front of your mind.

Vicki Fisher (25:47):
I kind of would like to know everybody's thoughts on just, I know you had mentioned earlier notes, which I think is really important because notes take up a lot of time. They get off the phone and they write the notes. You don't know if they're accurate or not. But for us, we use field adjusters for the field work and inside desk adjusters to finalize the claim. So are there any companies that are using anything to help pull all the data out of their independent adjusters report to make the inside job a little easier and a little faster? Is anybody doing anything with that? That's my next big challenge. So I am looking for your guys' ideas.

(26:26):
I know one of our vendors that we use who helps us with electronic payments recently came to us and said that they may start requiring some additional fields. And so that's part of what's we started thinking this is they're like, they want to know what rooms were damaged and it's so that the mortgage company can sign off on the payments and without us sending a full estimate over to the mortgage company. And it started me thinking, okay, how can I pull that data out of my documents? And they're not always coming in in the same format. And so that's one of the conversations we've been having is how can we use AI to read those documents and pull that data out for us so that we can then send it to our electronic partner, but also so that we could speed up the process of reviewing that report and settling that claim a little quicker for the insured.

Audience Member Dana (27:14):
What about virtual inspections? You guys utilize those? She's a technology vendor. Oh, I'm sorry, Kayla, go ahead. Go ahead. You can go.

Audience Member 5 (27:27):
I just want to say that a lot of these conversations is really surrounds around where your data is, absolutely whether your data is categorized, whether your data is ingested, whether you have a central data lake or data warehouse. So any of these things are already available, like the solutions are already available in the marketplace, but it's really more about setting up your data so you can have all these additional fields and then get a solution. Then you can try to improve the concept to see if it works for your organization.

Vicki Fisher (28:04):
That makes a lot of sense and it's something we're always working on is making sure our data is in the right format and in the right organization. And I think that's a challenge for every company as we have so much more data at our fingertips than we used to. A lot more data points. I know we look at our data points and there's millions of them and it's a lot to categorize,

Gina Hardy (28:24):
Oh, go ahead.

Tekeela Barnes (28:28):
One of the things we are as an organization is data is huge and one of our initiatives is building our own data warehouse and we're working on that now, trying to figure out what vendor to partner with because data is everything right about now.

Audience Member Dana (28:49):
Oh, okay. That's good.

Audience Member Emily Klo (28:52):
Hi everyone. Emily Klo with Allstate Right now I sit over health and benefits, but I have experience on the PNC side and partnership is so important. So if you don't have the partner and I look at my State Farm colleague over here, hi. But if you don't have the partner, there are so many, I think I'm a little close to the speaker, but there are so many

(29:21):
Data lakes out there that you can access through partnerships. So if you want to get the floor print or the floor plan of a home, there are a lot of technologies out there that vacuum the floor plans of homes and there are a lot of technologies and partners out there that have access to the way that people move through their life and it's all connected. So I would say at Allstate we always look to partner with folks that are emerging experts or experts in those fields so that we can tap into the way that people live and move through life and then become more proactive about how we can protect them. So if you don't have information or if you are building your data lake, being able to connect with partners I think is a way to truly expand that leaps and bounds.

Gina Hardy (30:18):
Wonderful, thanks. So what about some ideas? I know some of you are in technology and you're building stuff. What are you really excited about that you're building that for those of us that are carriers in the room, we should be excited about too. Okay, my deck, we have one

Vicki Fisher (30:33):
Minute, so we have one minute to wrap this up. So if you could give us your,

Audience Member Mike Pigeon (30:38):
Hey, I got the last, so if you're familiar with Duck Creek, we have a number of different products. We're a pretty large company, but one of the latest products is a product called Clarity and it's all about the data. So we partner up with Snowflake and in the whole idea there is you have to get your data curated and basically flying in the right direction and then you can do significant analytics on the back end of that. So it's not just Duck Creek that's doing this, the whole industry is moving towards this, but even outside the industry is moving towards it. But that's really kind of, if we want to talk about latest, greatest products, that's one of the latest products we have and it's just taking the level of data and what we can do with it to just a whole other extreme. And we could talk AI in the middle of that. So I think you got 30 seconds left so you can give it to someone else. 30

Gina Hardy (31:34):
Seconds breaking

Audience Member 7 (31:34):
Off. Yeah. Hey, so I'm with DXC technology and we compete with Duck Creek, but we're in the same state. But similar to that, putting your data, like you mentioned into a data lake, being able to then use tools to help you really analyze and create those reports that are going to help you all the way to your exposure in a certain area, how much you need to pay in claims, all those types of things. But it really is making sure that data is good and that's the key. If you have bad data, you're going to make bad decisions.

Vicki Fisher (32:21):
Thank you guys for all your feedback, we really appreciate it. Gina, would you like to add anything before we close?

Gina Hardy (32:26):
I was just going to say that this, we got asked to lead this discussion and it was certainly an honor leading it with each of you and certainly I hope that all of us came back with something from this discussion because certainly have a wonderful expertise in the room. So thank you guys.