Kyle Nakatsuji on What the Future of AI in Insurance Actually Looks Like

Kyle [0:00:00]: Many things connected to the rise of Ai, the advancement in Ai as pertains to, like, human capital and the economy will be k shaped.

Kyle [0:00:10]: Those who lean into it early and adopt early will see significant benefits and those who do not will see significant struggle.

Callan [0:00:20]: Welcome to the insurance growth lab.

Callan [0:00:22]: Who we go deep on the growth campaigns and strategies driving real results in the

Kyle [0:00:28]: insurance industry.

Callan [0:00:30]: Uptown Harrington founder flash growth and in each episode, I sit down with marketing and growth leaders from carriers, shirt tech and top to break down one specific initiative, whether it's highly marketed a product, scale a channel or solve a specific growth challenge.

Callan [0:00:45]: It's no fluff, just tactical insights you can apply in your own company.

Callan [0:00:50]: So, Kyle, there's probably...

Callan [0:00:56]: And I don't say this lightly?

Callan [0:00:59]: A hundred different directions I could have started this conversation out from my research.

Callan [0:01:04]: Where I'd love to start this is, can you tell us about trying to get into business school?

Kyle [0:01:11]: I can.

Kyle [0:01:12]: This is a fun story.

Kyle [0:01:13]: Although it reveals some things about me.

Kyle [0:01:15]: I'm not sure I should reveal, but whatever we'll go for it.

Kyle [0:01:18]: So I was a political science major and undergrad, which either means you care deeply about political science or you don't care about school.

Kyle [0:01:26]: And I was in the latter camp.

Kyle [0:01:30]: I was a football player and so that was an easy major to choose watched a lot of films on politics.

Kyle [0:01:34]: So I got on playing football and was forced into retirement due to skill.

Kyle [0:01:38]: And and never heard that.

Kyle [0:01:42]: That's that's how it works at the end of your career.

Callan [0:01:45]: You're retired.

Callan [0:01:45]: You're

Kyle [0:01:45]: not good enough keep playing.

Kyle [0:01:46]: So I had to find out what I wanted to do in the real world, and I obviously had given no thought to it whatsoever.

Kyle [0:01:51]: So decided to go to law school because that's something you can do if you were a political science major and you have no real career ambition.

Kyle [0:01:59]: So, like, I did went to law school and didn't know anything about being an attorney, but everyone seemed quite proud of me for making that decision.

Kyle [0:02:05]: So I did it.

Kyle [0:02:07]: And I got there, and we, a man with the university of Wisconsin.

Kyle [0:02:09]: A fantastic school.

Kyle [0:02:11]: Wonderful program.

Kyle [0:02:11]: I just mh.

Kyle [0:02:12]: Didn't realize after starting for six months.

Kyle [0:02:16]: I wasn't sure I wanted to be a lawyer.

Kyle [0:02:17]: Now that I knew what it was.

Kyle [0:02:18]: So the business schools across the street.

Kyle [0:02:20]: And I walked across the street and walked into the admissions office and said, I'm Kyle, I'm a law student.

Kyle [0:02:27]: I think I wanna go to business school too.

Kyle [0:02:29]: And they said, oh, okay.

Kyle [0:02:31]: So Do you have any experience in business?

Kyle [0:02:34]: I said no.

Kyle [0:02:35]: I've never done anything in business.

Kyle [0:02:37]: They said, well, what about, like, work experience?

Kyle [0:02:40]: Do you have any work experience?

Kyle [0:02:41]: I was like, I worked on baseball fields when I was in college, making sure they were ready for the game, but that's about it.

Kyle [0:02:47]: I was busy playing football.

Kyle [0:02:48]: And they said, alright, We'll have you taken the Gm map, and I said, why don't know what that is.

Kyle [0:02:53]: And and they said you can go, sir, what you don't?

Kyle [0:02:58]: You're gonna have a place for you here, which sounded a lot like a maybe.

Kyle [0:03:02]: And sure.

Callan [0:03:04]: That's what I would've taken from that.

Callan [0:03:05]: No question.

Kyle [0:03:06]: So I just started coming back.

Kyle [0:03:07]: Day after day, week after week was, like, a new plan of how they could let me in.

Kyle [0:03:13]: And the woman who worked in admissions at the time Aaron Nichols was such a saint she took those meetings.

Kyle [0:03:18]: And, like, finally, she convinced someone in Wisconsin Mba program, they should take me.

Kyle [0:03:25]: And so after I took the G twice.

Kyle [0:03:28]: I sort of wrote them letters and I met with people and, like, she convinced somebody, they should let me in.

Kyle [0:03:33]: So I got in to the business school at UW.

Kyle [0:03:36]: And it was the ultimate, like, moment when the dog catches the car.

Kyle [0:03:41]: Yep.

Kyle [0:03:42]: Where it was the only thing I wanted, and then I got it, and I was, what do I hell do I with this?

Kyle [0:03:46]: So I don't know anything about business.

Kyle [0:03:48]: So this is true.

Kyle [0:03:49]: I went to goodwill, and I bought a bunch of books from goodwill about business.

Kyle [0:03:53]: I bought, like, one up on Wall Street by Peter Lynch.

Kyle [0:03:56]: I bought this accounting book, which was oddly, like, it was, like long and thin.

Kyle [0:03:59]: It was like a weird book to hold on to.

Kyle [0:04:01]: It was very strange Picture book.

Kyle [0:04:03]: Yeah.

Callan [0:04:04]: For yeah.

Kyle [0:04:04]: Yeah.

Kyle [0:04:04]: Right.

Kyle [0:04:05]: Me I mean, I said accounting book.

Kyle [0:04:06]: It was a one two, three book for babies.

Kyle [0:04:08]: But, you know, let's leave that out of the story.

Kyle [0:04:10]: But I read.

Kyle [0:04:12]: And then I got there.

Kyle [0:04:13]: And promptly realized that I needed to figure out a job because they tell, like, now you're your physical, you what do you wanna do?

Kyle [0:04:19]: I, I know I just wanted to be here.

Kyle [0:04:20]: And they're, you gotta figure out a.

Kyle [0:04:22]: So I started talking to people, and I met someone who is in venture capital.

Kyle [0:04:25]: I went back to the admissions group, and I said, I wanna be in venture capital.

Kyle [0:04:28]: And they said you idiot.

Kyle [0:04:30]: It's like the only job you can't have.

Kyle [0:04:31]: You need, like, twenty years of work experience, like, you're here, and now you screwed it up again.

Kyle [0:04:36]: And so I also sounded like a maybe.

Kyle [0:04:39]: So I just stopped going to class.

Kyle [0:04:40]: And went out into the community to meet with startups and entrepreneurs and, like, get my hands dirty with investors and startups.

Kyle [0:04:47]: And ultimately got my degree and the program is great.

Kyle [0:04:50]: But, yeah, It was a bit of a winding journey to get there, decide that I didn't need to go to Class more go learn how to be an entrepreneur, go to class enough to graduate.

Kyle [0:04:59]: And now, stay very close to the program because, frankly, despite my absences, it very much gave me chances others would not and help sort of shape my knowledge and and what I brought to my career.

Kyle [0:05:11]: So I'm very appreciative for the opportunity.

Kyle [0:05:12]: It was just a bit of roundabout about journey.

Callan [0:05:15]: You know, what I hear that story.

Callan [0:05:17]: All that's going through my head is that's like the most...

Callan [0:05:20]: Founder story that you could possibly tell.

Callan [0:05:24]: Like, every single part of it.

Callan [0:05:27]: The no is really just a maybe the...

Callan [0:05:31]: I think I'm just gonna do it this way anyway.

Callan [0:05:33]: Like, you can't script a more founder story.

Callan [0:05:36]: So I love that.

Callan [0:05:38]: And you know what Here's the reality is, obviously, that ended up being a great decision, and that's paid off.

Callan [0:05:43]: You know, just talk about some of the stats when you've been able to accomplish, you know, raised over five hundred million dollars and funding a clear cover.

Callan [0:05:49]: And also staying on the same thing.

Callan [0:05:53]: The question that your boss asked you at Am fam.

Callan [0:05:58]: And I quote, so you wanna start a competitor in a market where there are top four leaders spend of combined six billion on advertising every year, and you want to spend barely any money on advertising to instead focus on experience.

Callan [0:06:13]: Before I even finish to highlight Real, When you sitting there, what was your response to that question?

Callan [0:06:20]: Well, I said yes.

Callan [0:06:21]: But

Kyle [0:06:24]: my response might have been different which just good what's going through my head, which is like, I think so.

Kyle [0:06:29]: But yeah.

Kyle [0:06:30]: You, we'll have to figure that out, I guess.

Kyle [0:06:33]: They were probably wiser an I.

Kyle [0:06:34]: Right?

Kyle [0:06:35]: Because I think we could have made some...

Kyle [0:06:36]: The thought was well placed.

Kyle [0:06:39]: We had no idea what we were getting into, but that's some of the beauty of entrepreneurship as well as because if you did, you wouldn't start That naive was ultimately turned into Hub bristol, which then ultimately turned into an education, and, hopefully, someday turns into success.

Callan [0:06:54]: Well, writing over a hundred and twenty three million in premium, the previous year, Here's probably one of the things.

Callan [0:06:59]: Of all the stats that are out, there and there's a million of these on, Clear cover and Dear born, which we'll talk about here shortly.

Callan [0:07:05]: You've recorded a seven minute claim payout.

Callan [0:07:09]: Mh.

Callan [0:07:10]: That's a pretty incredible number.

Callan [0:07:13]: When you consider the average forty four days, I believe and you hit this in seven minutes.

Callan [0:07:18]: How is that possible?

Kyle [0:07:21]: Great technology.

Kyle [0:07:22]: Team working closely with a claims team who knows what they're doing to figure out how to pull that together in a way that's safe and regulatory friendly and customer friendly.

Kyle [0:07:30]: So it was probably a lot of the things that now are the foundation of dear airborne labs.

Kyle [0:07:35]: Really?

Kyle [0:07:36]: If I'm being quite honest, we launched that at, like, two years ago, maybe longer.

Kyle [0:07:41]: And what the technology is capable of today blows that out of the water.

Kyle [0:07:45]: We just haven't worked on it just yet.

Kyle [0:07:48]: It was a cross functional collaboration between people who understand the insurance business and people are really good at building technology and that that is a massive unlock for any company.

Kyle [0:07:57]: And now with where technology has gone, that moment of collaboration can happen again, and again, and again, almost instantaneously because so much of the knowledge is now available to agents who can do that collaboration on your behalf and then surfaces it to humans who can make the final say.

Callan [0:08:15]: I wanna talk about this.

Callan [0:08:16]: You know, first why launched Dear airborne labs at all?

Callan [0:08:19]: You've got a successful company and calling it what it is?

Callan [0:08:23]: You're essentially giving competitors the tools to be successful.

Callan [0:08:26]: And especially where I've heard you state this and makes complete sense.

Callan [0:08:31]: Just so many of these carriers that are out there have been around for hundred years.

Callan [0:08:35]: I think both amp and nationwide or celebrating a hundred years this year.

Callan [0:08:39]: And to, you know, those are two of the bigger ones, but there's a ton of small ones that have done the same thing.

Callan [0:08:44]: The reality is with that type of technology, they should...

Callan [0:08:47]: I mean, they've got more data at anybody else.

Callan [0:08:49]: Why launch this at all?

Kyle [0:08:52]: It was a couple things I'd led to the decision to do this.

Kyle [0:08:54]: I would say that the decision to launch dear airborne labs took roughly ten years in eight weeks because it was ten years of hard won lessons building, spending a couple hundred million dollars just learning how to deploy production technology and Ai in a real insurance business with real insurance operators.

Kyle [0:09:14]: So I always spent a lot of time doing that and building the technology backbone, the assets, the experience you need.

Kyle [0:09:20]: It's And throughout the years, we had talked about whether or not we could monetize the technology in some way.

Kyle [0:09:28]: But to be candid, like, we didn't have a ton of interest in just selling software to people because that exists and people are very good at it and wasn't the business we really wanted to get into.

Kyle [0:09:37]: But we had thought about it.

Kyle [0:09:39]: And then roughly eight weeks ago, what the frontier models became capable of represented in our view a fundamental shift, a real transitional moment in terms of how knowledge work gets done.

Kyle [0:09:53]: And so much of insurance is individual knowledge work, we sort of crossed a threshold.

Kyle [0:09:59]: And the time for people to take advantage of what is possible is now and the time then for us would never be better to help people take advantage of that moment.

Kyle [0:10:10]: There was no longer any reason to wait because the technology had gotten to a place in the world had gotten to a place where you set it upfront.

Kyle [0:10:18]: If you are standing still you're falling behind, and you were falling behind at a faster rate than you ever have before because of how quickly things are moving.

Callan [0:10:27]: For me, it was the release of Opus four point six Claude Opus four point six and Chad Gp five point two.

Kyle [0:10:34]: Mh.

Callan [0:10:34]: And I know Chat Gp just launched at the time of this recording five four.

Callan [0:10:37]: But those two...

Callan [0:10:39]: And it was really five two that woke me up, and then I went back to Claude again, and I was like, oh my god.

Callan [0:10:44]: This Opus four point six is just on it I think it was just more than...

Callan [0:10:47]: And for me, it was intuition, and where that it was really like, oh, it's filling in the blanks, where it didn't do a great job.

Callan [0:10:54]: I mean and it did a good job, but now it's actively filling in those blanks and rounding out the story which felt like the missing piece.

Callan [0:11:02]: Now I know there's five million other things.

Callan [0:11:05]: But that's what it was for me.

Callan [0:11:06]: I'm curious what did you see

Kyle [0:11:08]: at that time?

Kyle [0:11:08]: Same.

Kyle [0:11:09]: I think with the new frontier models, the reasoning was substantially better.

Kyle [0:11:12]: So, like, the common refrain you'll here with folks is who've tried these things out six months ago.

Kyle [0:11:17]: Is they'll say, yeah.

Kyle [0:11:18]: They're just not that good.

Kyle [0:11:19]: It was wrong.

Kyle [0:11:20]: Couldn't do math.

Kyle [0:11:20]: Right?

Kyle [0:11:21]: And, like, that was all true to some extent, but I what model you're using six months ago.

Kyle [0:11:25]: The reasoning capabilities fundamentally shifted with Opus four point six.

Kyle [0:11:30]: And Saw looks pretty good.

Kyle [0:11:31]: But Opus four point six certainly, and then the new Open Ai models.

Kyle [0:11:34]: The reasoning suddenly took a big leap forward, there was almost no lag time between when the reasoning took a giant leap forward, and the ability for that agent to have access to context from all of these places where the context reside.

Kyle [0:11:49]: Yeah.

Kyle [0:11:50]: Also leap forward.

Kyle [0:11:51]: And that combination of things was this fundamental leap?

Kyle [0:11:55]: I built my own agent at the time just to play around with the technology.

Kyle [0:11:59]: And that was performative because it became obvious very, very quickly that if you are curious and proactive, and you know how to ask questions and you're persistent, you can build incredible things now.

Kyle [0:12:12]: Without having to know much else.

Kyle [0:12:14]: Again, that'll be a big unlock for knowledge workers.

Kyle [0:12:16]: Which for us was...

Kyle [0:12:17]: Look, there's no longer any time to wait if we're gonna help people the time is now.

Kyle [0:12:21]: Let's go do.

Callan [0:12:23]: Makes complete sense.

Callan [0:12:23]: Like, you're just using Claude c for the first time for me was...

Callan [0:12:27]: It's just like, okay.

Callan [0:12:29]: It's here.

Callan [0:12:29]: Agent ai, like, drew agent Ais here.

Callan [0:12:31]: Like, that was to me.

Callan [0:12:32]: It's very similar to first time Ever test drove.

Callan [0:12:35]: A tesla.

Callan [0:12:36]: First time, I ever test years ago, I was like, okay.

Callan [0:12:39]: Well, yeah.

Callan [0:12:40]: Self driving cars are here.

Callan [0:12:41]: It's just...

Callan [0:12:41]: There's a governor on.

Callan [0:12:42]: Like, I mean, like, there's limitations on what you can do, but it's here.

Callan [0:12:45]: Like, it's no longer a fig of people's imaginations, like, the things here.

Callan [0:12:50]: So an area that I wanna specifically focus on.

Callan [0:12:53]: I know we talked about the claims and all that matters as well for this as well, But I wanna focus on the agent experience.

Callan [0:12:58]: Mh.

Callan [0:12:58]: And you know, this is probably gonna be a mix of clear cover and dear airborne because I think a lot of this is probably driven by that same core technology.

Callan [0:13:06]: You know, one of the things that I hear all the time from agents and agencies is and I saw this a lot of penguin.

Callan [0:13:16]: More than anything than they want from care Of course, the market softening.

Callan [0:13:20]: So, like, a lot of big challenges that agencies experience with carriers are going to take care of themselves.

Callan [0:13:26]: The number one problem is going to take care of itself as the market soften.

Callan [0:13:29]: But the number two problem is ease of use.

Callan [0:13:33]: It's probably the most common thing that I hear personally next to appetite.

Callan [0:13:38]: How does what you're doing?

Callan [0:13:40]: And if you could share some examples on this from the clear cover side that'd be great.

Callan [0:13:44]: But how does what you're doing change ease of use onto the agency.

Kyle [0:13:49]: So clear, auto insurer writes primarily non standard are...

Kyle [0:13:53]: We distribute a bunch of different ways, but we do work with agents quite a bit while we launched the new non standard product in Florida recently that we're scaling out to agents in Florida now.

Kyle [0:14:03]: And so, like, we have a fair amount of experience in the agency's space working with these folks.

Kyle [0:14:07]: In dear more labs, it's like a hybrid sort of consulting slash technology build out business that, know I'm sort of remember pal here, but for mid market insurers.

Kyle [0:14:18]: And is targeted at carriers in Mg, but we've had a handful of conversations with agencies as well, because these are problems we've seen both sides of in our work at clear cover.

Kyle [0:14:27]: And so, like, I think both businesses sort of in some way touch this space.

Kyle [0:14:31]: So a couple of things.

Kyle [0:14:32]: And the refrain of, like, the agent ease of use.

Kyle [0:14:35]: The agents are busy they have a lot to do ease of use is always at the top of the list of any survey you put out the agents saying, like, who do you like and why do you like them?

Kyle [0:14:42]: When we launched Clear cover, we made a lot of investments in ease of use.

Kyle [0:14:45]: So how do we build software that's easy for the agent to engage with and looks good and, like, helps automate some tasks that they would normally have to do.

Kyle [0:14:53]: And so that was, like the old world of ease of use in my mind.

Kyle [0:14:57]: Where we're headed now, and we have not rolled us out to our agents in Florida, otherwise I suspect we will shortly is what constitutes an interface for ease of use is likely to change pretty substantially in the next couple of years.

Callan [0:15:10]: What's that look like?

Kyle [0:15:11]: So what we should all be doing, and I know Clerk ever will do.

Kyle [0:15:15]: I can't speak for everyone else is if I were an agent today, what I would want all my carriers to do is make anything I can do in their agent portable available to me via Mc.

Kyle [0:15:25]: What's that mean?

Kyle [0:15:25]: Available to me, essentially via a connection that an Ai agent can use.

Kyle [0:15:30]: Yep.

Kyle [0:15:30]: So that you can run your business through a set of Ai agents that you orchestrate that all have access to the tools, the data and the connections they need to do those portions of the job.

Kyle [0:15:42]: So like if you think about the roles within an agency, many of those things are tasks that could belong to an Ai agent.

Kyle [0:15:49]: But the ai agent cannot do the job unless the Ai agent has the context and the connection to the tool to do the job.

Kyle [0:15:55]: And so what I would be asking for as an agent, I'm sure we'll build this for them is in Mc, it's sort of like an Api, but for Ai agents.

Kyle [0:16:02]: And so if I'm the agent...

Kyle [0:16:04]: If I'm running an agency, I want to be able to work with my Ai orchestra who will orchestrate of series of other Ai workers who do specific tasks for me that are all connected up to Clear covers agent portal so that I know...

Kyle [0:16:18]: I don't have to go in there to get what I need.

Kyle [0:16:20]: It can all be surfaced up through my agents who can do the analysis and present me with it whatever I need to know about my business with Clear or any other carrier I work with.

Kyle [0:16:28]: My sense is that, this ease of use just like many other areas of software.

Kyle [0:16:32]: If you're an agent navigating seven to fifteen different carrier portals to do your job, plus your Ams, plus all the other things you need to do to run your business.

Kyle [0:16:41]: Those days are numbered.

Kyle [0:16:43]: Because all of those systems should be communicating with an agent and that agent should be your primary interface to do work.

Kyle [0:16:49]: Is not gonna happen overnight, but like that, again, like, the definition of ease of useful change substantially as this interface layer starts to fade away in favor of direct interaction with an agent who manages other agents on your behalf.

Callan [0:17:02]: Alright.

Callan [0:17:02]: We're gonna go on the weeds on this one because this is super interesting.

Callan [0:17:05]: Yeah.

Callan [0:17:05]: Okay.

Callan [0:17:06]: So if Fun playing that back.

Callan [0:17:08]: What I'm hearing you say is you're gonna have some sort of interface layer.

Callan [0:17:13]: This could be your Ams provider could provide this.

Callan [0:17:16]: It could be a third party that provides us...

Callan [0:17:18]: It could be connectors within one of the big platforms.

Callan [0:17:21]: Claude, Open Ai, Gemini, whatever that might be.

Callan [0:17:24]: And it sounds like this is pretty much up for grabs at this point.

Callan [0:17:27]: For the most part.

Callan [0:17:28]: I think I mean there are some mature text.

Callan [0:17:30]: I I know that are working on this, but nobody's got it down to what you had just said.

Callan [0:17:33]: And essentially, that is going to have this Mc p layer that will connect to all of the carriers.

Callan [0:17:41]: And whatever else.

Callan [0:17:42]: It could comb through your Ams, it can do...

Callan [0:17:45]: It's essentially just going to interact like an Api would, but it's built for agents with all these different platforms.

Callan [0:17:51]: And then either by rules that you've set up and how you like to do business with in the coverage amounts that you like is taking that against the duration just like you'd build a skill and Claude, that will serve up what the quoted.

Callan [0:18:04]: But everything else is gonna be done a hundred percent behind the scenes and the agents are gonna be doing all this.

Callan [0:18:09]: Is that essentially what you're saying?

Kyle [0:18:12]: Yes.

Kyle [0:18:12]: And it's even broader than that?

Kyle [0:18:14]: Because I think you're...

Kyle [0:18:15]: Again, how quickly we get there is a question of adoption.

Kyle [0:18:17]: And the technologies on its way, but some of it isn't all the way there yet.

Kyle [0:18:21]: But, again, at the pace that which things are moving, you can't bet it's that far away.

Kyle [0:18:24]: If you're running an agency when you walk in in the morning?

Kyle [0:18:28]: There's things running through your head that you need to accomplish today?

Kyle [0:18:32]: Why, I need to make sure that, like, are we closing business effectively?

Kyle [0:18:35]: Am I getting the top of funnel that I need if my retention good, Do I need to do outreach to anybody who's renewing soon?

Kyle [0:18:41]: Like, what's you have a task list and a whole bunch of other things that worry you?

Kyle [0:18:44]: Those are all things that you can just tell an agent to start for you?

Kyle [0:18:49]: To give you a first draft.

Kyle [0:18:51]: To take it as far as the agent you want the agent to take it before you ever have to think about it.

Kyle [0:18:55]: And so, like, you're an agent you walk in the door, like, I would love a world where all of Clear coverage agents can go talk to Claude.

Kyle [0:19:01]: Or whatever instance of, like, primary Ai interfaces is and say, which policy holder should I reach out to today?

Kyle [0:19:08]: And why?

Kyle [0:19:09]: And the work gets done in the background where you have an agent who then goes and looks at all the different agency portals that you have to manage and figures out which of your policy hilda are up for renewal shortly.

Kyle [0:19:18]: And which of them have churn before and, like, maybe are at risk of churn now.

Kyle [0:19:22]: And when's the last time I talked to them and that gets surfaced up to the orchestra agent who's managing all the sub agents who are going through the agency portal?

Kyle [0:19:29]: The orchestra agent now says, well, look, I have the list of people.

Kyle [0:19:32]: What do I know about these people?

Kyle [0:19:33]: Because I have, like, my little Crm over here and that tells me that, like, you know, Tom has this job.

Kyle [0:19:38]: So I probably shouldn't call them at this time.

Kyle [0:19:39]: I should call them this evening.

Kyle [0:19:40]: And, like that then gets serviced up the agent, and here's your top ten list.

Kyle [0:19:44]: The people that you should reach out to and here's when I suggest you do it.

Kyle [0:19:46]: And, by the way I drafted the communication for you because I know enough about Tom to make it sound personal.

Kyle [0:19:51]: Yeah.

Kyle [0:19:52]: We are not that far away from that being real.

Kyle [0:19:54]: What we need to enable it for the agent is these connections.

Kyle [0:19:57]: The oil in this new world is context.

Kyle [0:19:59]: Pure symbol.

Kyle [0:20:01]: And the agents will need context available and mass and unified for the agents tend to go and do this work on it math.

Callan [0:20:11]: Who do you think gonna make that happen.

Callan [0:20:14]: Like, because the carriers are a part of this.

Callan [0:20:16]: Right?

Callan [0:20:16]: Like, the carriers are a part and can play their part with that.

Callan [0:20:20]: And typically, carriers have a lot of the resources to be able to invest in things like this.

Callan [0:20:25]: Is this gonna be the Ver four applied, xi waves?

Callan [0:20:29]: Is this gonna be more of the policy admin systems that are going to unlock some of these things like, who do you think?

Callan [0:20:36]: Or is it multiple players?

Callan [0:20:37]: What does that look like in your mind?

Kyle [0:20:39]: Yeah.

Kyle [0:20:39]: Probably all of the above?

Kyle [0:20:40]: So one of the things that we have worked on at, Clear cover, and I think it'll be important in Dear airborne labs engagements.

Kyle [0:20:47]: Our goal is we're a service provider first.

Kyle [0:20:49]: We come in and we wanna help people understand what the world looks like what their opportunity set looks like and what to execute on first us.

Kyle [0:20:57]: And we think we're well positioned to do it because we've done this for a decade.

Kyle [0:21:00]: We've solved a bunch of these hard insurance and technology problems for a decade.

Kyle [0:21:03]: So we come in, and you identify that as a carrier, like, you wanna a claims c copilot.

Kyle [0:21:08]: Because that claims copilot is gonna help your reps manage three times more claims and they can handle right now because they're gonna do a bunch of work on the reps behalf and then bring it back to them for judgment and curation and taste and all the things that the reps still needs to be involved with.

Kyle [0:21:22]: Great.

Kyle [0:21:22]: So we can build that for you, I have one.

Kyle [0:21:25]: We ran it our own business, a hundred percent of our reps use our claims copilot.

Kyle [0:21:29]: We're happy to build that technology for you.

Kyle [0:21:31]: But it turns out that if you plan on doing it again, and again, What you should probably do is build this middle wear layer.

Kyle [0:21:39]: We call our context graph, which is essentially the thing that uni identifies all the data in your business, and then layers that context on top of it.

Kyle [0:21:46]: So the next agent you build not only has access to whatever data that person who is doing that test before has, but it has access to what the claims agent is doing too.

Kyle [0:21:56]: Because we all know if we build you an underwriting agent, it would sure be nice if it knew exactly what the claims agent knew one was doing, and they could collaborate together.

Kyle [0:22:03]: Insurance very cross functional business for agents to work cross functionally, they need shared context.

Kyle [0:22:07]: If you need you need shared context, you need a shared platform.

Kyle [0:22:10]: And so for us, building that context graph, which is the unification layer and then a pending of context that this is a map of the business, and all the things that are in it and what they do and who they are and what their are, very, very important if you wanna scale this up.

Kyle [0:22:25]: And agents will be no different.

Kyle [0:22:26]: There will probably be any number of people who continue to serve an important back office function, their software, but their interface will be less important.

Kyle [0:22:34]: They will be feeding up into a middle wear layer, something that collects all of the data that resides in this agency, and that the agents can live on top of and do the job directed of them by the orchestra, which is ultimately the person running the business.

Kyle [0:22:49]: And so, like that, data aggregation and context curation layer.

Kyle [0:22:53]: I don't know who will build that for agencies.

Kyle [0:22:55]: Maybe we will, wouldn't be against it.

Kyle [0:22:57]: But Yeah.

Kyle [0:22:58]: That's the thing that I think needs to exist.

Kyle [0:23:00]: And then everyone else still exists, but you're interface with the customer changes because someone is gonna own that, like, primary agent interface.

Callan [0:23:11]: Super interesting, and it makes complete sense what you're saying because it's only as good as what you can feed it.

Callan [0:23:17]: And I know that, like, that was what it was before.

Callan [0:23:20]: So I think it still is the case now.

Callan [0:23:22]: And you know, an area where I'd like to dive into on this, and a very similar note is you guys had a clear covered rain survey.

Callan [0:23:29]: This was in twenty twenty one.

Callan [0:23:31]: And with four hundred seventy five agents and ninety eight percent said that digital tools made their jobs easier, but seventy four percent also fear being replaced by technology.

Callan [0:23:44]: I bet those numbers.

Callan [0:23:45]: I bet they haven't changed much at all.

Callan [0:23:48]: Right.

Callan [0:23:49]: Maybe went up in some cases.

Callan [0:23:50]: You know, when you hear that as a carrier, how do you support an agency through that?

Callan [0:23:57]: Like, what are your thoughts on that?

Callan [0:23:58]: Is it what you're saying?

Callan [0:23:59]: Like, we need to build this piece for them for them to be successful?

Callan [0:24:03]: I think there's a lot...

Callan [0:24:05]: Of course, the insurer that was released, you know, the Ai agent connection with chat Gp sent shockwave waves everywhere.

Callan [0:24:11]: Mh.

Callan [0:24:12]: Not to say it's not a cool app, but I don't think it's fully there yet.

Callan [0:24:15]: Right.

Callan [0:24:15]: But I'd like to get your thoughts on that.

Callan [0:24:17]: And just in general.

Kyle [0:24:18]: I think it's still very much true.

Kyle [0:24:19]: And I will not deny that I think many things connected to the rise of Ai, the advancement in Ai as pertains to, like, human capital and the economy will be k shaped.

Kyle [0:24:33]: Which is to say, like, those who lean into it early and adopt early, will see significant benefits, and those who do not will see significant struggle.

Kyle [0:24:42]: Those two lines will diver emerge.

Kyle [0:24:43]: And so, like I think that is very real.

Kyle [0:24:45]: But in the agency, space in particular, agents still need help with today.

Kyle [0:24:50]: Humans with all their capabilities.

Kyle [0:24:52]: There's a couple things that are very human native.

Kyle [0:24:55]: But first is relationships.

Kyle [0:24:57]: As it stands like the agents are very good at sounding human and they can make videos that look like humans and like, we get closer every day.

Kyle [0:25:04]: But relationship still matter in this business.

Kyle [0:25:06]: And so the agent as a relationship manager for the firm that has a very large team of agents who are doing much of the work that they had to do before, still super important.

Kyle [0:25:16]: It's still relationship business that's still very much matters.

Kyle [0:25:19]: The agent will have to continue to manage that.

Kyle [0:25:21]: The other thing that agents are not great at yet, is c their own context.

Callan [0:25:26]: Agents being Ai agent.

Callan [0:25:28]: Ai agents agency.

Callan [0:25:28]: Yeah.

Kyle [0:25:29]: Yeah So the Ai agent isn't back great at creating its own context.

Kyle [0:25:32]: And so in one of the roles of the human agent in this process is making sure that the Ai agents they have working for them have access to all of the context they need to do the job well.

Kyle [0:25:42]: Because, again the agent's is very smart.

Kyle [0:25:43]: It's pretty much limited at this point by the context you can give it.

Kyle [0:25:46]: And so I think the human agent as the, like, the orchestra and context curator for the Ai agents that work for them, and that's an important human job it will continue to be Like, I...

Kyle [0:25:55]: You know, this thing I know this thing is important for the agent to know to do its job better.

Kyle [0:26:00]: I have to make sure it hasn't it.

Kyle [0:26:02]: I have to make sure that it's relevant.

Kyle [0:26:04]: I have to make sure it stays updated that context, like, they call it harness engineering in the tech world, which is how do you make sure the agents are sort of architect and then given context in a way that allows them to be most effective.

Kyle [0:26:18]: That is a human job still the human agent will do that.

Kyle [0:26:21]: And then the third is taste, I think that there's still an element of a human looking at an output and deciding if it is sort of crafted to elicit the response or achieve the objective it is intended to.

Kyle [0:26:38]: And the ai gets better at that every day too, But, like, there is still some important element of taste, which help making decisions on, like, what something ultimately looks and feels like with respect to the experience you're giving to your customer.

Callan [0:26:53]: Yeah.

Callan [0:26:53]: Here's a piece of accountability too.

Callan [0:26:55]: Right?

Callan [0:26:55]: People want to know that somebody is accountable to this decision.

Callan [0:26:59]: And that's when I think it's gonna be really interesting.

Callan [0:27:02]: And I've heard this point get brought up, and it's one that I've kinda thought about is if you...

Callan [0:27:07]: I'll just use Chad, but it could be any.

Callan [0:27:09]: And you go through and you fully buying your policy, and you've got a claim and care and denies said claim.

Callan [0:27:16]: Where's that liability lock?

Callan [0:27:18]: Mh.

Callan [0:27:19]: That is an interesting one.

Callan [0:27:21]: Is that going to be it carriers and open Ai?

Callan [0:27:24]: Gonna be, like, suing each other back and forth.

Callan [0:27:27]: Where does that lie?

Callan [0:27:28]: That's an interesting one.

Callan [0:27:29]: Not saying it's not sol in some way shape or form, but I think people just in general want to know that somebody else.

Callan [0:27:36]: That could change.

Callan [0:27:37]: You know, another area I'd love to circle back on is the k shaped.

Kyle [0:27:42]: Mh.

Callan [0:27:42]: And I think that's very interesting.

Callan [0:27:43]: And I tend to agree with it.

Callan [0:27:46]: I mean, honestly.

Callan [0:27:47]: For me, I think of, like, I said this yesterday other day I was like, look, this thing's probably gonna take my job here at some point, but it is it's all fun to mess with until that point.

Kyle [0:27:57]: Yeah.

Kyle [0:27:57]: Yeah.

Kyle [0:27:57]: Let's you.

Callan [0:27:58]: So we're recording right now on March twelve?

Callan [0:27:59]: I know this won't be released for a couple of weeks and this could change in a couple of weeks.

Callan [0:28:03]: But when you think of that k shape and, you know, the people that are leading into it now, where do those people need to be?

Callan [0:28:10]: And I'll give you like, some examples.

Callan [0:28:12]: Does that mean that They just kinda have to use Ai as far as, like, they're asking it questions using as a thought partner.

Callan [0:28:19]: Do they need to be...

Callan [0:28:21]: Let's just go, like, middle the road using c to start to execute on some tasks or let's go to the top where they've got five mac minis and open claw agents running on all the different machines?

Callan [0:28:34]: Like, in your eyes, like, what is leaning into it right now on that case shape look like?

Kyle [0:28:39]: Just starting, just doing anything.

Kyle [0:28:41]: The time is now.

Kyle [0:28:43]: And if you're not doing something you are absolutely falling behind, but you don't need to be an expert.

Kyle [0:28:47]: And I think this is a big trap because I think people...

Kyle [0:28:49]: When they get started, like, they wanna craft the perfect prompt.

Kyle [0:28:52]: They wanna know exactly what to tell the agent so that it wow them.

Kyle [0:28:56]: And that's not actually how it works.

Kyle [0:28:57]: You just have to start.

Kyle [0:28:58]: I'll give you a tangible example here.

Kyle [0:29:00]: So, again, like, I built my own assistant in cloud code and Jarvis And the Jarvis is eight weeks old.

Kyle [0:29:06]: Fantastic tool, like, we built a brain, and we built a routing infrastructure for how it acts this is the parts of its brain and a memory and learning for certain we've got all this stuff, which sounds complicated and that I supposed to make it sound complicated now.

Kyle [0:29:19]: But it started because I read an x post tweet.

Kyle [0:29:23]: From somebody who said, I built myself an Ai assistant and cloud code.

Kyle [0:29:28]: And I took the link, and I put it in Claude, and I said I want one.

Kyle [0:29:32]: Can you build this?

Kyle [0:29:34]: And it said, sure.

Speaker_2 [0:29:36]: Here's how I would do it.

Speaker_2 [0:29:37]: And I was Okay.

Kyle [0:29:40]: Get started.

Kyle [0:29:40]: And now you're sort of tumbling down the rabbit hole, but you don't need to really know

Callan [0:29:45]: how to start.

Callan [0:29:45]: You just have to start.

Kyle [0:29:46]: Would you need the prerequisites to success in my mind in this new, like, agent capability world is curiosity.

Kyle [0:29:53]: Yeah.

Kyle [0:29:54]: Like, being able to ask questions, understand when something is interesting and press on it.

Kyle [0:29:58]: And then prerogative, taking those questions and saying, like, I'm just gonna keep pushing on this with my agent until I get the outcome that I'm seeking or I learned something that changes my direction, but like, with the tools available now, like, you don't have to know how to do anything.

Kyle [0:30:13]: You just have to know how to ask to admit it.

Kyle [0:30:15]: And then ask how to do it, and you can figure it out.

Kyle [0:30:18]: You don't have to have five open call.

Kyle [0:30:21]: That would be cool, and those people are ahead.

Kyle [0:30:23]: But you don't have to have five open call agents running your whole life in order to be caught up at this point.

Kyle [0:30:28]: You just have to start.

Callan [0:30:31]: Yeah.

Callan [0:30:31]: You've got a quote out there.

Callan [0:30:33]: And I heard you say I said a conference before.

Callan [0:30:35]: What I remember hearing that at the conference like Already, I gotta get him on the show.

Callan [0:30:38]: This was last year's Short tech insights in New York.

Callan [0:30:41]: And you told the crowd, you said, if you were a little behind before, you're going to be really behind in six months.

Callan [0:30:48]: And I think that was the one that really set in on everybody.

Callan [0:30:52]: Mh.

Callan [0:30:52]: When do you feel that that's gonna slow down that statement.

Kyle [0:30:57]: One thing I'm gonna take away from this is I better watch what I say on stage.

Kyle [0:31:00]: Because I'm I wasn't sure anyone was paying attention, but now

Callan [0:31:03]: that you've tracked it.

Callan [0:31:04]: I gotta be a little careful.

Kyle [0:31:05]: Thank you for listening.

Speaker_2 [0:31:06]: So I'm not sure it's gonna get better ever.

Kyle [0:31:12]: Technology does have been in waves and, like, we are certainly at a point where things are happening fast.

Kyle [0:31:17]: Every single day, new product, new skill, new feature.

Kyle [0:31:21]: This this this.

Kyle [0:31:22]: You can do this with cloud code.

Kyle [0:31:23]: You can do this with Open Ai.

Kyle [0:31:24]: We're certainly at a pretty tall a steep part of the curve.

Kyle [0:31:28]: But the whole thing about Ai when it hits that inflection point, which again, like, I think we probably did just hit is that the curve becomes self reinforcing.

Kyle [0:31:37]: And the smarter the tools get, the smarter the tools are able to make the next set of tools that you need.

Kyle [0:31:43]: The faster the tool.

Kyle [0:31:45]: And and so, like, we're probably my opinion is we're at the point where the curve is unlikely to get much flatter ever.

Kyle [0:31:50]: Because the tools are getting increasingly good at creating the next tables

Callan [0:31:54]: starting to hit that exponential curve.

Callan [0:31:55]: You know, one of the questions I'm curious.

Callan [0:31:58]: We talked a lot about agent experience Ai just in general.

Callan [0:32:01]: Let's say you're talking to one of these mutual.

Callan [0:32:04]: And a lot of the older mutual many of them are a hundred percent dependent on independent agents.

Callan [0:32:09]: And they come to you guys and what's the first piece of advice you're giving them if they say, I wanna improve my agent experience.

Kyle [0:32:18]: To be honest, the first thing we're gonna do is ask a bunch of questions.

Kyle [0:32:21]: And trying and understand what do you like about it now?

Kyle [0:32:24]: What do you think the shortcomings of it now?

Kyle [0:32:26]: Like, have you talked to the agents recently about where they feel like you have opportunities to improve?

Kyle [0:32:31]: But does the tech tech look like now and it does it limit you in any way?

Kyle [0:32:34]: Like, there's a host of things that I'd have to learn to make a good recommendation because I think that...

Kyle [0:32:40]: This is the thing that I...

Kyle [0:32:41]: My hope is that this separates Dear one labs from some of the other things that are out there.

Kyle [0:32:46]: Is it there's a ton of great software out there.

Kyle [0:32:48]: But it begins and ends at its own walls.

Kyle [0:32:51]: And in many cases, a hammer in search of a nail.

Kyle [0:32:55]: We have hammers too.

Kyle [0:32:56]: We build a whole bunch of it over the last decade.

Kyle [0:32:58]: But the way we wanna go to market and help people is by first understanding the landscape of what they do and what's possible and what works and what doesn't, and then figuring out how to deploy the right hammer to the right nail at the right time to make sure that the house gets built appropriately and as quickly as possible.

Kyle [0:33:17]: So We're gonna ask a bunch of things just to try and understand, like, what is the core problem?

Kyle [0:33:22]: You are trying to solve here and why.

Kyle [0:33:24]: From there then, I think the the next part of the answer once we understand that is How do we gather as much context as possible around this interaction?

Kyle [0:33:36]: This task.

Kyle [0:33:37]: The people involved, the things involved, the actions that those people and things take and make that available to an Ai who can help us all get smarter around how this ought to work.

Kyle [0:33:48]: And that, like, this collecting of context in order to have Ai help guide us in the right direction is critically important, and it turns out that it is really an advantage of incumbents as it relates to using Ai because the incumbents have lots of context.

Kyle [0:34:06]: Sometimes they can't access it.

Kyle [0:34:08]: Sometimes it resides in people's head instead of a knowledge base or it's not directly accessible and there's some work done to really draw it out, but carrier that's gonna around for a hundred years has a lot of context.

Kyle [0:34:17]: They have a lot of lessons learned and knowledge and they have collected a lot of data and, like, that context is again of soil in this world, like, unlocking that and making use of it can be a real competitive advantage for a company that's been around a long time.

Callan [0:34:30]: What is that discovery process?

Callan [0:34:31]: I'm assuming it's a discovery process?

Callan [0:34:33]: Is that, you know, interviewing the territory managers, interviewing their top agencies interviewing all of the people, both executives, honestly, probably everybody is my guess.

Callan [0:34:45]: Yep.

Callan [0:34:46]: To get all that information, is it surveys?

Callan [0:34:48]: Is it d all the above?

Callan [0:34:50]: What does that look like?

Kyle [0:34:51]: It can be any of those.

Kyle [0:34:52]: And we also have built some tools internally at clear cover that help us accelerate this process where you basically give Ai access to different pockets of information and then the Ai can scan and interpret and analyze and surface what it thinks the big opportunities are you as well.

Kyle [0:35:09]: It's like we built some technology to help with this discovery process.

Kyle [0:35:12]: To be honest, like, we're not professional consultants.

Kyle [0:35:16]: We're operators.

Kyle [0:35:17]: We've done this for a decade and have learned hard lessons.

Kyle [0:35:20]: And so my plan isn't to go in was like, a cookie cutter or deck for every one of these discovery engagements.

Kyle [0:35:25]: Our plan is to, like, get there, understand the issues.

Kyle [0:35:30]: Use what tools we have at our disposal and the things we've done at Clear to try and figure out where to direct our efforts and resources.

Kyle [0:35:36]: And then work with the client to really understand the promise space and what that solution is.

Kyle [0:35:41]: So, like, we can do the consulting thing.

Kyle [0:35:43]: And I think there's value in that and maybe we partner on it.

Kyle [0:35:45]: But what's important is that we work with the client in whatever way makes sense to get to a good decision on what to build to maximize financial return for their business.

Kyle [0:35:56]: And this we learned is less than the

Callan [0:35:58]: hard way at clear because we built a

Kyle [0:35:59]: lot of stuff that was interesting, but had to really start to understand how to build the organizational Dna around prioritizing for impact on the business.

Kyle [0:36:10]: And that's not a slide deck.

Kyle [0:36:12]: That's just like we've experienced it, and we know where to look and how to help you look in the right places to take this thing you wanna build and translate it to a business outcome that we're both seeking.

Callan [0:36:22]: You said this earlier and it's definitely been my experience, it's really fun making the hammer.

Callan [0:36:28]: Like, when you could see the capabilities that these tools can do, and I can't even imagine.

Callan [0:36:32]: Right?

Callan [0:36:32]: With the team of engineers that are super supercharged.

Callan [0:36:36]: The hammers that they can build, but you can waste a lot of time.

Callan [0:36:40]: Mh.

Callan [0:36:40]: Just building hammers in taking that time to understand what are the right hammers to build.

Callan [0:36:45]: And I think this applies to carriers, agencies, it doesn't matter because there's so many cool stuff in All gonna be things you wanna do.

Callan [0:36:51]: But you can waste a lot of time and resources by selecting the wrong hammer.

Callan [0:36:56]: So I think that's super interesting.

Callan [0:36:57]: Kyle, last question that I have for you is tie this full circle, if you can have a conversation You know what, let's say with the same kyle that was leaving law school to walk over to try to get his Mba.

Callan [0:37:15]: What would that conversation be?

Callan [0:37:17]: What advice would you give them?

Kyle [0:37:20]: That's a good question.

Kyle [0:37:20]: Walk faster?

Callan [0:37:23]: Walk

Kyle [0:37:24]: You know, like, all of the stumbles in there have been many.

Kyle [0:37:27]: All of these sort of things that you would look at, retrospectively.

Kyle [0:37:31]: The things that you would change are usually the things that you're were like, man, that was difficult.

Kyle [0:37:35]: And I didn't like the feeling of that experience, but those things are formative.

Kyle [0:37:40]: You wouldn't give those up.

Kyle [0:37:42]: If you thought there was some risk that they changed where you ultimately landed.

Kyle [0:37:46]: So like, there's a million things that were difficult to go through or that I made difficult to other people, and I have regret over making those things difficult for other people.

Kyle [0:37:56]: But the lessons make you who you are.

Kyle [0:37:59]: So, yeah.

Kyle [0:38:00]: If anything, like, just go faster, try more.

Kyle [0:38:03]: Do more...

Kyle [0:38:04]: Like, I would have packed more learning into the last fifteen years if I could because that's what has made all difference.

Callan [0:38:12]: It's interesting.

Callan [0:38:12]: And it is saying, like, I find, when you make those major mistakes, that's when actually the true change happens.

Kyle [0:38:20]: We have made many, and we've hopefully learned from all of them.

Kyle [0:38:24]: And some of those we've now turned into a business.

Kyle [0:38:26]: That, you know, has, like, accumulation of lessons learned and mistakes made in the form of advice and technology.

Kyle [0:38:33]: But, yeah, We made any of them.

Kyle [0:38:35]: But I don't know that I really regret any because, like, they all were formative and they'll lead you to a place where you're better for them as long as you'd let them do that.

Callan [0:38:44]: I love it.

Callan [0:38:44]: Kyle, that's a perfect place to bring this to a wrap.

Callan [0:38:47]: I appreciate coming on today.

Callan [0:38:48]: I think this is super interesting.

Callan [0:38:50]: I think there's so much opportunity for those like you said that are on that k right now and leaning in.

Callan [0:38:56]: So I appreciate you breaking everything down today.

Kyle [0:38:58]: Thanks for the invite Callan.

Kyle [0:39:00]: Absolutely.