Latest / Breaking the ICE / Episode 15 - Demystifying Enterprise Insurance Technology and how AI Shapes the Future

Transcript

Episode 15 - Demystifying Enterprise Insurance Technology and how AI Shapes the Future · Breaking the ICE

• 0:02
  SPEAKER_01: Hello and welcome back to the Breaking the Ice Podcast.
• 0:06
  SPEAKER_01: This is uh season one, episode 15.
• 0:09
  SPEAKER_01: Uh I'm joined by my co-host, Lloyd Overton.
• 0:12
  SPEAKER_01: Lloyd, how are you doing today?
• 0:13
  SPEAKER_01: Yeah, not bad idea.
• 0:15
  SPEAKER_00: How are you?
• 0:15
  SPEAKER_00: Looking forward to the weekend.
• 0:17
  SPEAKER_01: Yeah, yeah.
• 0:18
  SPEAKER_01: It's uh it's an exciting weekend here.
• 0:21
  SPEAKER_01: It's the Toronto Marathon weekend.
• 0:23
  SPEAKER_01: Uh cool.
• 0:24
  SPEAKER_01: So yeah, yeah, yeah, yeah.
• 0:26
  SPEAKER_01: Yeah.
• 0:26
  SPEAKER_01: Oh, you did it?
• 0:27
  SPEAKER_01: You're doing it.
• 0:28
  SPEAKER_01: So you you you've taken part?
• 0:30
  SPEAKER_01: You know, I was meant to, uh, and unfortunately got a got a been fighting like a chest infection.
• 0:37
  SPEAKER_01: So uh doctor's orders were to not do it.
• 0:40
  SPEAKER_01: Um same thing happened a couple of years ago, and uh I ended up just doing one on my own, just around my neighborhood, uh just to kind of uh, you know, before the end of the year, I kind of, you know, once I'm I'm good and and and back to uh back to full health, I'll I'll go and do one on my own.
• 1:00
  SPEAKER_00: It's kind of like marathon season though, isn't it?
• 1:02
  SPEAKER_00: Because I I um I know they did London in like April or whatever, but we went away last weekend to um to Ghent and Bruges, and uh when we got to Bruges, we was like, God, this is so difficult to get into the city at.
• 1:14
  SPEAKER_00: I know it's small and quite pedestrianized and stuff like that, but we didn't realize it was the Bruges marathon, so we they all seem to be going on at once, which is cool.
• 1:20
  SPEAKER_01: Yeah, Chicago was last weekend too.
• 1:22
  SPEAKER_01: Spring and fall um are are kind of the season for it to keep the temperatures down.
• 1:27
  SPEAKER_01: Yeah, for sure.
• 1:28
  SPEAKER_01: Although two years ago, uh it was about 25 degrees Celsius without the sort of humidity uh in Toronto, uh and it was horrendous.
• 1:38
  SPEAKER_01: I mean, I was like, you don't you don't plan for that weather and and you dress for the cold, and it just uh the temperature went up so quickly.
• 1:48
  SPEAKER_01: And the back half of the Toronto marathon is all along Lake Ontario, the just right along the waterfront.
• 1:54
  SPEAKER_01: And um uh it's it's so there's no there's no protection from just brutal, brutal sunlight.
• 2:02
  SPEAKER_01: Uh so it was, yeah, it was a sweaty one for sure.
• 2:05
  SPEAKER_00: Problem is you don't really um like in in the you can I've done London once, I've only done one marathon in in 2013, and and you don't train for it either because you know in London you're training all over the winter when it's cold and wet and rainy.
• 2:15
  SPEAKER_00: And then if you do rock up in April and and actually it's you know sun's out and it's quite warm, you're like, oh god, you know, have I really got to do this for 26 miles?
• 2:23
  SPEAKER_00: Yeah.
• 2:24
  SPEAKER_01: But I will be going down to support all the rest of the racers this weekend, so uh I'll be down there Sunday morning.
• 2:29
  SPEAKER_01: Yeah, yeah, yeah.
• 2:30
  SPEAKER_01: That'll be fun.
• 2:31
  SPEAKER_01: Yeah, yeah.
• 2:32
  SPEAKER_01: What's what's new and exciting on your side of the Atlantic Ocean?
• 2:35
  SPEAKER_00: What's new and exciting on my side of the Atlantic Ocean?
• 2:37
  SPEAKER_00: So there's no there's definitely no marathons or anything like that that I'll be attending this weekend.
• 2:41
  SPEAKER_00: It'll be uh another weekend of unfortunately nursing the kids who have all got the logi.
• 2:45
  SPEAKER_00: You've probably seen the news and that there's a razor covid or some other kind of variant flying around everywhere.
• 2:51
  SPEAKER_00: So yeah, be a bit of a nursemaid there.
• 2:53
  SPEAKER_00: And I think, you know, uh typically I seem to uh seem to be kind of warming up to something as we get into the weekend as well.
• 2:58
  SPEAKER_00: But uh yeah, and then from a word perspective, it's you know, obviously off over the the weekend last weekend with a bit of a long one.
• 3:06
  SPEAKER_00: But yeah, we had a we had a good session in London with uh with the prospect talking about some uh some interesting stuff in the in the commercial space again, which is where ICE is is certainly kind of seeing a bit of momentum and and some of the the sorts of things that we've uh we've been doing recently and and clients one is kind of lending ourselves to to having more meaningful conversations in that space.
• 3:24
  SPEAKER_00: So that was a good session that could work out well for us.
• 3:27
  SPEAKER_00: But yeah, it's just it's just crazy busy.
• 3:29
  SPEAKER_00: There's loads going on.
• 3:30
  SPEAKER_00: It's like new financial year for us, as you know.
• 3:33
  SPEAKER_00: Um so it's all about right, how do we how do we get off and uh how do we get off to a fast start and and and make a real positive impact on on this next year?
• 3:41
  SPEAKER_01: Yeah, it is crazy busy.
• 3:44
  SPEAKER_01: Uh the past two and a half, and you know, usually there's a dip in the summer.
• 3:48
  SPEAKER_01: There wasn't this year.
• 3:49
  SPEAKER_01: Uh I mean, July, August were far busier than past years.
• 3:54
  SPEAKER_01: September was insane, and now October, November.
• 3:59
  SPEAKER_01: I mean, I don't think there's five minutes to spare, to be honest, Lloyd.
• 4:03
  SPEAKER_01: So I think that's what we wanted to roll with this episode, kind of go off the cuff, because you and I, you and I are out there uh having tons of conversations with folks.
• 4:14
  SPEAKER_01: And uh, you know, it's it's funny how there are still I I would argue some some funny misconceptions in our space of of you know enterprise technology.
• 4:26
  SPEAKER_01: I wanted to talk through them, sort of demystify some of uh some of these misconceptions.
• 4:31
  SPEAKER_01: You know, we sell an enterprise platform, and I think that one of the biggest set of common misconceptions is what's involved in selling and implementing enterprise tech.
• 4:46
  SPEAKER_01: Uh, and whether that's from a time standpoint, a cost standpoint, a um uh you know, uh a process standpoint, uh, you know, I think there is this air of um complexity and high cost and large timescales, you know, we have to fight back against and and you know, we have to do it through case studies, testimonials, uh references, uh showing our work basically, right?
• 5:19
  SPEAKER_01: You know, as much transparency as possible in the process to to it to to highlight, you know, that it doesn't necessarily have to be that way.
• 5:27
  SPEAKER_01: Um, but yeah, maybe, maybe that's where we can start.
• 5:30
  SPEAKER_01: Um let's try to break down the the the illusion uh of of time and cost and complexity uh related to to making such a big change.
• 5:40
  SPEAKER_00: Yeah, no, it's uh it's a good point.
• 5:42
  SPEAKER_00: And and it's one of I've spoken about it a few times this week, just in in relation to some of the kind of prospects and people I'm talking to.
• 5:48
  SPEAKER_00: You know, I I wish it was like selling a lottery ticket, I really do, you know, but it's just not like that.
• 5:52
  SPEAKER_00: You know, it takes a lot of effort.
• 5:54
  SPEAKER_00: It takes a lot of work.
• 5:56
  SPEAKER_00: Um and and again, you know, I think it's it's right that it takes a lot of work, and it's it's right that it takes a lot of effort because you know, people aren't buying a lottery ticket.
• 6:04
  SPEAKER_00: That's the point.
• 6:05
  SPEAKER_00: They're buying uh, you know, a transformational, quite often, piece of IT that is going to be put into a you know, we've done a lot of large stuff recently, but again, could be large, small or medium, but it's going into a business that is either being set up for growth to become a big business and really make a difference, or a big business that's very much established, that's got a lot of people in there that want to want to have a kind of improvement to their day-to-day life and how they interact with our software.
• 6:29
  SPEAKER_00: So, you know, it's a big decision.
• 6:31
  SPEAKER_00: And and you know, let's face it, let's not be around the bush here.
• 6:33
  SPEAKER_00: It's not cheap.
• 6:34
  SPEAKER_00: You know, you need you you're making a big commitment um financially and a big commitment to to making that change.
• 6:39
  SPEAKER_00: So, you know, quite rightly so, these insurers have got to go through lots of gates to make sure they're comfortable with that decision.
• 6:45
  SPEAKER_00: And and and I think, you know, for me, having kind of been with ICE for for for eight and a half years and seeing it from a smallish business to now becoming a you know a big player, certainly with the wins we've had in the tier one space, some of the biggest ways and most convincing ways that we've been able to kind of get our point across on what we're doing is just through it's through just proof, right?
• 7:06
  SPEAKER_00: You know, I I remember when we were really pushing for a tier one and we didn't necessarily have a tier one, and then you know, we got a big brand in John Lewis, so all of a sudden you've got you've got a bit more credibility in that space.
• 7:16
  SPEAKER_00: And then all of a sudden we we were kind of bridesmaid on a couple, but then we bagged RSA, and then you know, again, that just paved the way for for a viva.
• 7:24
  SPEAKER_00: So again, you know, it's it makes it so much easier when you can go into these kind of sales situations and say, look, you know, don't believe us.
• 7:30
  SPEAKER_00: Have a look at this.
• 7:31
  SPEAKER_00: You know, here's the here's the proof, or you know, give us a give us a problem and we'll solve it for you.
• 7:36
  SPEAKER_00: Um so yeah, you know, I think I think these things get easier as time goes on, but not uh not getting away from the fact it can be pretty long in terms of uh sales cycles.
• 7:46
  SPEAKER_01: Yeah, and and let's unpack some of those things, right?
• 7:49
  SPEAKER_01: Because I I think you know there was a lot there um that you alluded to.
• 7:54
  SPEAKER_01: I I guess first and foremost, I think one of the biggest contributing factors to uh you know managing these big enterprise transformations um more smoothly is how much more standardized our products and services have become as our company has evolved over time.
• 8:21
  SPEAKER_01: Right.
• 8:21
  SPEAKER_01: So as a and and I and I think you know what what um what insurers find when they're working with relatively new companies is that because the product is less refined, uh there's so much more customization that needs to happen in order to get it to be fit for purpose for them.
• 8:42
  SPEAKER_01: We're very mature enterprise product suite on the balance of things, right?
• 8:48
  SPEAKER_01: I mean, we're talking about 20, 25 years in in market and in industry.
• 8:52
  SPEAKER_01: So we're so so much farther along the product lifecycle, um, very feature rich.
• 8:58
  SPEAKER_01: Uh so much comes out of the box, off the shelf, you know, configuration, not code.
• 9:04
  SPEAKER_01: Uh and because of that, it it I think that's one of the most embedded uh sort of project accelerators when we partner up with new companies, right?
• 9:14
  SPEAKER_01: Um so the more that you can accomplish through that sort of standard off-the-shelf offering, as opposed to having to, you know, gather requirements, build technical specifications, you know, engineer against those specifications, test, and then worry about that new function or feature breaking something else in the in the tech stack, you know, we we mitigate a ton of project risk, a ton of project cost, and and save time, you know, by by not having to go down that path.
• 9:47
  SPEAKER_01: Uh, you know, we've been doing a few projects lately with absolutely zero customization, implementing just the core product, right?
• 9:54
  SPEAKER_01: And obviously there's there, you know, we sit within an ecosystem.
• 9:58
  SPEAKER_01: We're a core platform, so there's technical integrations and there's there's you know entry and exit points from the confines of our system to the the the broader tech stack, but within a within the box, you know, within our our single product, um, we're not having to to write new code.
• 10:18
  SPEAKER_01: Uh you know, so so that's it's a huge, a huge um simplifier.
• 10:25
  SPEAKER_00: Um yeah.
• 10:26
  SPEAKER_00: I think it's it's a mindset thing as well, isn't it?
• 10:28
  SPEAKER_00: You know, it's it's kind of sharing those experiences perhaps with those customers that we're working with and kind of convincing them that, you know, guys, you don't need to boil the ocean here.
• 10:37
  SPEAKER_00: You know, I think actually kind of delivery from a project perspective has definitely changed in certainly the period I've been working in insurance from you know very long, elongated, you know, project processes to a kind of agile sprint-based delivery where you know you're saying, look, at the end of this phase, I want to see this.
• 10:53
  SPEAKER_00: At the end of this phase, I want to see this.
• 10:54
  SPEAKER_00: It's kind of breaking it down and in into kind of its component parts and seeing progress right the way through rather than the saying, look, we start here and we finish here.
• 11:02
  SPEAKER_00: I think I think it has made a big difference.
• 11:04
  SPEAKER_00: But again, I think, you know, again, going back to your point around the type of systems and platforms that we're sending, actually, you know, let's get let's get some meaningful benefit quickly.
• 11:14
  SPEAKER_00: Let's get this done.
• 11:15
  SPEAKER_00: Let's get you out there, let's get you trading, let's get you writing GWP.
• 11:18
  SPEAKER_00: Ultimately, that's what an insurer is is attempting to do.
• 11:21
  SPEAKER_00: And then from that point, we can we can bolts and bells and whistles on around the side, we can do the next bit in phase two, we can do the next bit in phase three.
• 11:28
  SPEAKER_00: Actually, you might even think to yourself, phase three is a totally different phase than what I thought it was, because actually maybe we don't need some of these things.
• 11:35
  SPEAKER_00: So it's kind of, you know, it's a it's an education piece and bringing people on a journey that you know we've seen successfully kind of put into place with others.
• 11:43
  SPEAKER_01: Yeah, I I I think um one of the biggest historical concerns with the phased approach versus sort of the big bang approach is that there was going to be a lot of rework when when you're doing a phased rollout.
• 11:55
  SPEAKER_01: Um, that there's certain steps in the process that just become multiplicative because uh they're they're required steps anytime you do a deployment.
• 12:06
  SPEAKER_01: Um what I'll say is uh you know, I think we have a really, really strong migration team and migration strategy.
• 12:15
  SPEAKER_01: And so that lends itself uh that lends itself well to reducing the sort of the multiplicative steps um involved with a phase rollout.
• 12:27
  SPEAKER_01: And so then again, that feeds into start small or pick an MVP or you know pick a logical phase one.
• 12:36
  SPEAKER_01: Maybe it has to do with you know, I I found it in in you know our half of the business, Lloyd, that oftentimes customers are consolidating multiple different platforms onto us as a single operating system.
• 12:54
  SPEAKER_01: And so that phased rollout often has to do with like the contractual obligations for the sunseted systems, right?
• 13:02
  SPEAKER_01: You know, this one is running off next year, okay.
• 13:05
  SPEAKER_01: Well then that's where we're gonna start, right?
• 13:07
  SPEAKER_01: And then the next one runs off in two years, that'll be phase two.
• 13:10
  SPEAKER_01: Um and again, I mean, it it just the flexibility uh of our of our platform and ease of use uh you know supports that sort of um uh face rollout as well.
• 13:25
  SPEAKER_00: Yeah, no, I agree with that.
• 13:26
  SPEAKER_00: And and the migration question is an interesting one because you know, coming from an insurance background myself, I can remember that you know people wincing on the other side of the table when you mentioned something called migration, they're like, oh God, no, I don't want to go down that path.
• 13:37
  SPEAKER_00: And I can remember kind of arriving at ICE, who had been through a few migrations with clients at that point, some big bang and and then some at renewal.
• 13:44
  SPEAKER_00: And you start talking to people that that that kind of been through it from a technical standpoint, and you're like, right, you know, oh no, but hang on a minute, this isn't like migrating 30,000 um, you know, policies.
• 13:53
  SPEAKER_00: We're talking like three million here.
• 13:55
  SPEAKER_00: And they're like, well, no, that's not really the complexity.
• 13:57
  SPEAKER_00: You know, the whether it's three thousand or three million, if you get the if you get it right up front in terms of you know how you're extracting it, what you're putting, how you're aligning the data, what how you're preparing it to be kind of loaded into a new system, that's where the effort is.
• 14:10
  SPEAKER_00: And and and you'd need to go through that process, whether it's 30,000 or 3 million.
• 14:14
  SPEAKER_00: So actually being able to to kind of express that capability from a nice perspective, and our migration team do a cracking job at that is yeah, is is another one that kind of definitely demystifies something that a lot of people are pretty um pretty friend of when it's part of a project.
• 14:29
  SPEAKER_01: Yeah, I I I I wonder too if uh the this the shift toward uh cloud infrastructure and then the corresponding shift toward you know our customers and target market having their own sort of data lakes has also supported that.
• 14:48
  SPEAKER_01: Because you know, historically you would have been taking, or you would have like all of these enterprise platforms would have been operating in isolation, the data would have been uh you know, so much more segregated from the rest of the data that the rest of the platforms contained.
• 15:05
  SPEAKER_01: Now everything is streamed into data lakes, uh, or or you know, data warehouses.
• 15:11
  SPEAKER_01: And there's so whether it's fabric or snowflake or you know, pick pick your uh pick your uh you know data tool uh du jour, but that's so much more embedded into our customer strategies.
• 15:25
  SPEAKER_01: And so because of that, our role as a data processor in the full tech stack is very different, right?
• 15:33
  SPEAKER_01: All we're fundamentally doing is just plugging into one of these services and streaming to them, and so then because that's the underlying infrastructure, uh, migration becomes a lot simpler as well, right?
• 15:46
  SPEAKER_01: Um, you're not necessarily concerned about how you know data object A is gonna fit with B C D and E.
• 15:54
  SPEAKER_01: B C D and E are already spoken for because they're streaming into the same service as well.
• 15:59
  SPEAKER_00: Definitely.
• 16:00
  SPEAKER_00: No, I think you're right.
• 16:01
  SPEAKER_00: And and then, you know, let us overlay AI on top of that, you know, you're not writing scripts to transform it from that to this and move it from here to there before you get it ready to load into a new platform.
• 16:10
  SPEAKER_00: Actually, you're you're telling an AI service to to do some of that for you.
• 16:14
  SPEAKER_00: And you know, I know uh again from an ICE perspective, with use we've used that approach with a recent client.
• 16:19
  SPEAKER_01: Yeah.
• 16:20
  SPEAKER_01: What uh Wade, what's what's your take on the sort of being prescriptive in uh in in the process versus being reactive?
• 16:32
  SPEAKER_01: Um you know, every every customer says they want us to come in and and you know pitch our best practice way of doing things.
• 16:40
  SPEAKER_01: Um and very you know, that that always reaches uh reaches some resistance at some point in the chain.
• 16:47
  SPEAKER_00: Yeah, that's a that's a good question.
• 16:49
  SPEAKER_00: I I think it depends it depends on the audience and the insurer, right?
• 16:54
  SPEAKER_00: I think that that's definitely something to consider.
• 16:56
  SPEAKER_00: Um but I mean, you know, I I think actually, you know, the whole discovery phase for me is is such an important part actually of an implementation because you're almost getting as part of that process the ground rules in place as to how we do this.
• 17:10
  SPEAKER_00: So I know it that I know that it's a phase to kind of you know, meet and greet, how are you?
• 17:14
  SPEAKER_00: This is us, this is you.
• 17:15
  SPEAKER_00: We'll build out a project plan, we'll look at our gaps, we'll refine any estimates, and and that's absolutely what the primary function of it is.
• 17:21
  SPEAKER_00: But actually, in that in that first few weeks, you you can define how you want to work.
• 17:25
  SPEAKER_00: How does it work for you guys?
• 17:26
  SPEAKER_00: Who's your decision makers?
• 17:28
  SPEAKER_00: Who are your SMEs?
• 17:29
  SPEAKER_00: How are we gonna, as a project, make sure that this is a success?
• 17:32
  SPEAKER_00: So I think having that phase up front, you can almost kind of dictate what that looks like and agree it between you before you then go into that implementation phase.
• 17:41
  SPEAKER_00: So for me, you know, again, having having not necessarily seen that working for an insurer and previously, and then learning, right, what is a discovery?
• 17:49
  SPEAKER_00: How does it work?
• 17:50
  SPEAKER_00: Sitting in on a few of them and and watching them, super powerful, you know, piece of work to do up front.
• 17:55
  SPEAKER_00: It really is.
• 17:56
  SPEAKER_00: Um so for me, I'd kind of flip it on its head and say actually, I'd be using that phase to kind of work out what that approach looks like rather than you know, is it going to be prescriptive, reactive, or actually, you know, are we just gonna work, you know, uh and adopt a bit of both, really?
• 18:12
  SPEAKER_01: Yeah, I yeah, absolutely.
• 18:14
  SPEAKER_01: Um I think just the overall um the overall coming to terms on governance, communication protocols, uh, you know, how are teams going to work with one another?
• 18:28
  SPEAKER_01: Um I I think you know what matters and what doesn't matter.
• 18:31
  SPEAKER_01: And and so a common set of uh or a common understanding on priorities.
• 18:36
  SPEAKER_01: Um, you know, where are the pain points?
• 18:38
  SPEAKER_01: What what are the quick wins or or what are the the sort of the the biggest sort of burning elements um that need to be fixed right away?
• 18:47
  SPEAKER_01: Uh, you know, and and getting to that common set of priorities is is really, really important.
• 18:53
  SPEAKER_00: Definitely.
• 18:53
  SPEAKER_00: And I and you know, I was thinking about something that um you know came up as a as a bit of a discussion point around, you know, tier ones are really slow.
• 19:02
  SPEAKER_00: Um, you know, well, actually, I I don't buy that.
• 19:05
  SPEAKER_00: You know, I I think you know they can be more complicated, and absolutely there's some more complexity and speed aren't, you know, well, should I say complexity and speed aren't necessarily the same thing.
• 19:15
  SPEAKER_00: And I think actually when a tier one really gets behind something, they can move incredibly quickly because they've got lots of people, lots of expertise and lots of investment available to make that happen.
• 19:24
  SPEAKER_00: And I think again, making sure that that is a is is a kind of success point comes back to the exact thing you've just said, which is you know, when you've got those people in the room, and there tends to be a lot in those kind of scenarios, it's where where do we need to get to, guys?
• 19:37
  SPEAKER_00: What are the pain points?
• 19:38
  SPEAKER_00: What are the key priorities?
• 19:39
  SPEAKER_00: Because that's what we want to blaze through really quickly, and you guys are gonna be the ones to help us get there, and this is how you're gonna do it.
• 19:46
  SPEAKER_01: Yeah, and and I do find, I mean, you know, customers have structured their operations in a logical way.
• 19:51
  SPEAKER_01: We structured our platform in a logical way.
• 19:54
  SPEAKER_01: It may not fit like a glove uh immediately, but but I think that's where if you can prioritize the the biggest sort of burning issues, oftentimes in solving for that, the the the sort of the the rest of the the logical design of our platform, of our core processes, they help to then uh inform or instruct how that sort of process re-engineering should should take place, right?
• 20:23
  SPEAKER_01: Like let's solve these three things.
• 20:28
  SPEAKER_01: And in solving those three things, it's gonna set out the playbook or the blueprint for how the rest of the 15 things you know should then eventually fit.
• 20:38
  SPEAKER_00: Agreed, definitely.
• 20:40
  SPEAKER_00: You know, and we've we've seen that, right?
• 20:41
  SPEAKER_00: You know, we've we've seen that so many times.
• 20:43
  SPEAKER_00: And it just comes up to just getting yourself geared up and set up right from the very start, and and how that discovery phase can be so powerful when making sure that happens.
• 20:53
  SPEAKER_00: Yeah.
• 20:53
  SPEAKER_01: Um let's pivot.
• 20:55
  SPEAKER_01: I you know, I I was having a conversation with somebody yesterday, and we were talking about uh of course we were talking about AI.
• 21:04
  SPEAKER_01: Um and I I'm I'd like to get your take, and and and I'd I'd obviously like to share you know some snippets from the conversation I was having in terms of how AI is um maturing into the industries that we're supporting.
• 21:25
  SPEAKER_01: I guess that's maybe the the best way to put it, right?
• 21:27
  SPEAKER_01: So, you know, if you think about sort of B2B to C and working back from the consumer standpoint all the way back to you know our our product and and the products and services that that we're providing, um, what have you been hearing and seeing lately?
• 21:44
  SPEAKER_00: Do you know I I had a call with uh with a CTO today, right?
• 21:47
  SPEAKER_00: And we were talking about different things, and obviously AI came up in the conversation.
• 21:51
  SPEAKER_00: It comes up in every conversation now.
• 21:53
  SPEAKER_00: And I was talking a bit about what we're doing in Alice and our suite of services that we're offering to insurers and and to them specifically.
• 22:00
  SPEAKER_00: Actually, and and she then kind of got onto the point of talking about AI individually and what they're doing inside their business.
• 22:06
  SPEAKER_00: And I was like, We're doing the same, guys.
• 22:08
  SPEAKER_00: You know, we've got these agents that are kind of talking to our PMs and BAs and processing things from a project perspective.
• 22:14
  SPEAKER_00: We've got the engineers using stuff to kind of um interrogate code and stuff like that.
• 22:18
  SPEAKER_00: Um, and and I just kind of had a chat with her about some of the stuff we've done around kind of pricing and approach there.
• 22:24
  SPEAKER_00: And she said, Yeah, it's funny, right?
• 22:25
  SPEAKER_00: Everybody's now a software engineer.
• 22:27
  SPEAKER_00: And I'm like, do you know what?
• 22:28
  SPEAKER_00: You're right.
• 22:29
  SPEAKER_00: You know, actually, you know, everybody now has some level of ability when it comes to utilizing AI to be a software engineer.
• 22:36
  SPEAKER_00: I think there's there's this narrative that's kind of still going round that, you know, eventually AI just makes a decision and takes the person out of the process.
• 22:44
  SPEAKER_00: And, you know, do I think that's that's kind of where it ends up?
• 22:47
  SPEAKER_00: You know, I don't.
• 22:48
  SPEAKER_00: I I think from my experience, AI is brilliant at at helping people deal with information and summarising things and extracting data and taking some of that kind of repetitive work away.
• 22:59
  SPEAKER_00: But you know, insurance especially, there's there's a massive amount of kind of um independent and individual judgment in it, right?
• 23:06
  SPEAKER_00: You know, an under an underwriter looking at a complex commercial risk or a claims handler dealing with a um, you know, a sensitive um claim, you know, isn't isn't necessarily processing data.
• 23:18
  SPEAKER_00: There, you know, there's the there's context and experience and and judgment involved.
• 23:23
  SPEAKER_00: So I think, you know, the opportunity the opportunity still for me is you know, is AI making the human better and faster?
• 23:30
  SPEAKER_00: That's kind of uh naturally I think there will be some areas where AI does take take some jobs, but that's kind of, you know, I'm trying to be the optimist here and think about it that way.
• 23:39
  SPEAKER_00: You know, I just think we're getting too hung up on whether AI replaces somebody, and I think the more interesting question is whether it can make that person dramatically more effective.
• 23:47
  SPEAKER_00: It's definitely making me more effective, and it's definitely definitely helping me, and I'm embracing it from that perspective really.
• 23:54
  SPEAKER_01: So the the the the take that I sort of you know had yesterday was in in it in its in its uh most base form of description, AI is just sort of probability.
• 24:11
  SPEAKER_01: It's it's it's just sort of a probabilistic model that uh is is going to take the context it has to uh and try to provide an answer or an outcome based on that context.
• 24:25
  SPEAKER_01: And uh obviously I think that there's a bit of misconception around uh construing generative AI as all AI.
• 24:34
  SPEAKER_01: There are many, many, many different industrial applications of of AI that have been in practice for years, if not decades, right?
• 24:45
  SPEAKER_01: I mean, you know, uh we have customers that have process automation, that have um even you know machine learning built into some of their um data analytics processes or document uh document analysis, right?
• 25:00
  SPEAKER_01: Like all of those are subsets of AI.
• 25:02
  SPEAKER_01: AI is always doing the same thing, right?
• 25:05
  SPEAKER_01: It's it's interpreting data and then it's making a guess based on uh the algorithms that are programmed.
• 25:13
  SPEAKER_01: Generative AI, it's just that the guess that it makes is to generate something, whether it's uh text or a picture or video now or audio.
• 25:23
  SPEAKER_01: I just read a hilarious article about some guy who's been selling AI generated songs on Spotify, and and I think he he made like nine million dollars over the last decade, and and now he's he's been sort of caught and and arrested and had to give it all back.
• 25:39
  SPEAKER_01: But uh I guess it just tells you all you need to know about the state of modern music.
• 25:43
  SPEAKER_01: Um that's a combination conversation for another day.
• 25:46
  SPEAKER_01: Uh but anyways, so you know inherent in that, I think that if we can sort of decouple a few of these things, um generative AI is really, really good at generating.
• 26:02
  SPEAKER_01: And I think we're all sort of experiencing that on a consumer standpoint now.
• 26:06
  SPEAKER_01: You you see emails that have just you know inflated in size, you see LinkedIn posts that are now essays, you see, you know, uh, you know, I mean, even just you know, business-to-business communication has just gone off the rails because generative because people are using generative AI without guardrails.
• 26:26
  SPEAKER_01: The point that, you know, and and if you've been following along for the past couple of years, my tagline is that context is now king.
• 26:35
  SPEAKER_01: So the the off-the-shelf generative AI tools are trained on the entire corpus of of text, which means that they're they're hyper-generalized.
• 26:48
  SPEAKER_01: They're not contextualized to any, you know, one specific domain.
• 26:54
  SPEAKER_01: And so then you can't you can't rely on that um generative output to provide a contextualized response to your business, your industry, your you know, domain of expertise.
• 27:09
  SPEAKER_01: What makes our brains different from the AI brain?
• 27:15
  SPEAKER_01: You know, the generative AI brain is is gonna have a much broader set of knowledge and understanding of everything in the history of written word, but it won't necessarily know the uh the correct context to use that breadth of knowledge and understanding in, whereas our brains will.
• 27:38
  SPEAKER_01: You can get it there, but you have to contextualize it.
• 27:42
  SPEAKER_01: You have to provide it with the necessary data inputs to train it on whatever you know specific uh you know output you want it to become a master of.
• 27:57
  SPEAKER_01: And so I think as of right now, um the training contextualization of models, you know, building of agents and and training them is where focus should be applied, right?
• 28:10
  SPEAKER_01: And it's you know, we're not only doing it within the confines of our uh business, but also in in the products that we're now building, right?
• 28:21
  SPEAKER_01: This is this is where our area of focus is.
• 28:24
  SPEAKER_01: Um things like B2B co-pilots, things like you know, internal agents, they're only going to become useful when they've uh developed enough certainty uh based on the data they've been trained on and and the training processes that have been implemented to reduce the probability of incorrect output, you know, you know, from from again, just you know, an off-the-shelf generalized output, right?
• 28:56
  SPEAKER_01: That because we serve the insurance industry, that risk propensity is fractionally smaller when you then put it out to the consumer.
• 29:07
  SPEAKER_01: Right?
• 29:08
  SPEAKER_01: So you know, is is any insurer going to um you know, risk certain types of consumer interactions to a probabilistic model?
• 29:21
  SPEAKER_01: Um off the shelf?
• 29:24
  SPEAKER_01: No.
• 29:25
  SPEAKER_01: As the training of these models uh and the contextualization of them gets better, possibly yes, probably yes, right?
• 29:33
  SPEAKER_01: Um and and that's and that's just irregardless or or sort of sort of uh you know, that's that's without considering the need for human interaction or the the the place of of human to human interaction in uh sort of you know operating flow, um that that that's that's just considering the risk that you're willing to take in using a model to produce the right or correct answer for you.
• 30:00
  SPEAKER_01: Um so I think you know the biggest area of focus for me in getting from where we are today to a more uh AI first op you know operating model is is how well uh companies can contextualize um you know the agents that they want to deploy.
• 30:19
  SPEAKER_01: And and I I I wonder I wonder whether or not uh you know companies have sufficient data to be able to do that.
• 30:28
  SPEAKER_01: Right?
• 30:29
  SPEAKER_01: Or or or or is it is it going to become more of an industrial solution to tr to you know collect and and train third-party tools on a broader data set in order to make them more likely to uh come up with the correct solution to problems.
• 30:49
  SPEAKER_00: Yeah, I think it's a good point.
• 30:50
  SPEAKER_00: Uh, you know, and I and I think you frame it well in terms of you know context rather than just you know this broad range of information that might not necessarily be as as focused as you need it to be.
• 31:01
  SPEAKER_00: I had a conversation with somebody again today, actually, and they were kind of saying, you know, utilising AI from an engineering perspective, um, you know, how how does that impact kind of new developers that are coming on graduate programs or they're coming straight out of school and stuff and they're immediately just kind of grabbing hold of Claude and helping it write code, you know, how it's probably a good thing, and it's probably making it more effective and more efficient, but how do you know what what what processes are in place from from you know the point you make to ensure that it's being done right?
• 31:33
  SPEAKER_00: So, you know, is that code that's being written that's 16 lines, could it have been done in six?
• 31:38
  SPEAKER_00: How do we know?
• 31:39
  SPEAKER_00: Because and how do they know?
• 31:40
  SPEAKER_00: Because they're just using it and and adding their knowledge on top of that and thinking that's right, whereas someone with experience who can really kind of go down that route of of the context of how we do it, go, well, actually, no, you shouldn't have done those things.
• 31:52
  SPEAKER_00: You're absolutely right to embrace it and work with it, but think about it like this.
• 31:56
  SPEAKER_01: It's a remarkably powerful tool, uh, in the right hands, and and with the right set of operating uh uh operating controls around it, right?
• 32:07
  SPEAKER_01: I mean, I I think you know, we are certainly entering a new industrial revolution from the standpoint of how powerful um uh you know modern AI tools are in you know rewriting a lot of the knowledge economy.
• 32:27
  SPEAKER_01: But you know, provided that the correct uh control mechanisms and guardrails are are put in place.
• 32:34
  SPEAKER_01: Um otherwise there's gonna be a lot of waste and a lot of inefficiency that that that becomes as a byproduct of this as well, right?
• 32:41
  SPEAKER_01: Um I think a great analogy for this, um, you know, I have young kids who are are in the uh the the education system now, and there's a big risk, right?
• 32:54
  SPEAKER_01: That you know kids are gonna outsource learning to AI now, right?
• 32:58
  SPEAKER_01: Uh rather than learning two plus two.
• 33:01
  SPEAKER_01: Um they're just gonna ask what two plus two is.
• 33:04
  SPEAKER_01: And that's obviously a risk.
• 33:06
  SPEAKER_01: That's obviously a risk.
• 33:07
  SPEAKER_01: But again, if if you think about the potential with guardrails, um you know, rather than going to a a general tool, you know, having um having a curated tool where you can't ask two plus two, but you can ask it to explain why two plus two equals four.
• 33:24
  SPEAKER_01: Or you can ask it to uh, hey, you know, I need to I need to work on uh okay, addition is a pretty uh simple example.
• 33:33
  SPEAKER_01: You know, I need to work on long division, you know, create 20 sample problems for me and then walk through the solution.
• 33:42
  SPEAKER_01: Like that, so I mean, the ability to harness these tools for you know productive good is is remarkable.
• 33:50
  SPEAKER_01: Um but but again, the appropriate controls and guardrails need to be put on them so that you know people aren't just outsourcing uh everything to them w without any filter and and outputting the kind of slop that we see that's run rampant today.
• 34:05
  SPEAKER_00: Yeah, I'm kind of uh I'm totally with you on that.
• 34:08
  SPEAKER_00: And and I'll again it's another thing I've been thinking about quite a lot recently because my my kids are a bit older than yours, I think, Elliot.
• 34:14
  SPEAKER_00: And and Harrison is he's going into his GCSE year, he's in year 11, and you know, I'm still looking over the shoulder and seeing kind of Snapchat AI helping him with some homework questions, and I'm kind of like now saying to him, Look, I get it, you're using it, because there's no getting away from it, but you know, just think about how you're using it.
• 34:31
  SPEAKER_00: You're now in a year where you know these exam uh papers are going to be put in front of you without the use of AI and you're gonna need to come up with an answer.
• 34:37
  SPEAKER_00: So, you know, you should be asking it questions in a way that it's explaining to you how it's got to that answer.
• 34:42
  SPEAKER_00: But I think to your point, if you know we can we can build agents around actually within the AI that that kind of takes that approach right from the outset, that's the way that we should be using it rather than just you know at the end of the day, we don't want to bring up a generation of dummies, do we, that can only do anything if they're uh if they're kind of interrogating an AI cap an AI platform.
• 35:02
  SPEAKER_01: Yeah, and at the same time, we don't want to roll back the clock and have everybody counting on Abacai, right?
• 35:08
  SPEAKER_01: So I, you know, I like obviously progress is good.
• 35:12
  SPEAKER_01: Um and and yeah, again, I mean I lived through the I lived through this the transition from physical encyclopedia to digital encyclopedia.
• 35:24
  SPEAKER_01: And you know, how many people got caught red-handed just sourcing early Wikipedia where you know anybody can edit anything?
• 35:33
  SPEAKER_01: And and and so again, that's where you know the control mechanism in that in that circumstance was that somebody who's actually going to read the sources, somebody who's actually going to read the paper and grade it, you know.
• 35:46
  SPEAKER_01: Um and and and so it's the same thing here, right?
• 35:48
  SPEAKER_01: We just need to come up with with with what those uh you know processes and control mechanisms are, like you said, you know, I don't I don't think the education system is going to move away from uh you know manually written exams and manually graded exams um for that exact reason.
• 36:07
  SPEAKER_01: Now, you know, transitioning that back to the conversation we were having about industry speed support, that's that's kind of the concept of human in the loop, right?
• 36:15
  SPEAKER_01: It's just ensuring that there are stage gates and and uh you know opportunities and milestones for people to intervene and make sure things are um being output correctly, uh, you know, and and and and then they can they can proceed.
• 36:34
  SPEAKER_00: I was thinking about it another way only only uh the other day.
• 36:36
  SPEAKER_00: So, you know, I'm talking more and more now about oh, how did that mop test go?
• 36:39
  SPEAKER_00: Oh well done, mate.
• 36:40
  SPEAKER_00: You know, it sounds like you're making good progress there and stuff.
• 36:43
  SPEAKER_00: And I said to him, you know, have have you got anywhere where you feel that you know you need a bit of extra help or you feel there's any weaknesses, or there's some stuff you don't quite understand, or you're a bit you know, concerned about what grade you might get or where you're at at this point in time.
• 36:54
  SPEAKER_00: And he said, No, no, you know, I'm feeling pretty good about it.
• 36:56
  SPEAKER_00: But in my head, I was kind of thinking, you know, actually, you know, ten years ago you would have gone, right, okay, your maths is not quite where it needs to be.
• 37:02
  SPEAKER_00: We'll get you a tutor for an hour every week.
• 37:04
  SPEAKER_00: You know, you could probably build an agent that that can come up with some some you know, come up with an hour program that they can sit through once a week or twice a week or whatever it is, and do almost the same thing.
• 37:15
  SPEAKER_00: I mean, you know, I'm not saying that, you know, there's no need for tutors anymore, but you get my point.
• 37:19
  SPEAKER_00: It's you know it's utilizing it for the right reasons and thinking about it differently.
• 37:23
  SPEAKER_01: Yeah.
• 37:23
  SPEAKER_01: And and and so again, now, you know, think about you know creating uh training agents from a corporate training standpoint.
• 37:32
  SPEAKER_01: Yeah right.
• 37:32
  SPEAKER_01: Think about the potential um, you know, to really enhance uh you know individual um training and development, learning and development, you know, on an individualized basis in a corporate setting.
• 37:46
  SPEAKER_01: It's remarkable.
• 37:47
  SPEAKER_01: Um, you know, the the potential for uh individualization.
• 37:53
  SPEAKER_01: Um but again, I mean, these sort of B2B applications are you know they're they're much more closer to reality and and you know, in many instances they're already they've already been live and running because the stakes are lower.
• 38:11
  SPEAKER_01: Right?
• 38:12
  SPEAKER_01: Uh you know, if if uh if one of those internal models goes awry, uh the stakes are much, much lower than in the sort of contractual relationship between, you know, company and customer.
• 38:24
  SPEAKER_00: Agreed, agreed, especially when you're talking to you know global tier one blue chip companies.
• 38:29
  SPEAKER_00: Yeah, there's uh there's some pretty um there's some pretty big consequences of getting it wrong.
• 38:34
  SPEAKER_01: Um But there is a ton of excitement on the horizon, and and again, just you know, maybe uh Barry or maybe just you know foreshadowing what's on our roadmap, you know, I think we we do absolutely see the future is gonna be internal copilots, internal agents uh augmenting the um operators relationship with our system through the use of of a copilot, right?
• 39:04
  SPEAKER_01: So um, you know, that's gonna be some something that they interact with on a on a real-time basis, uh, you know, based on the experiences that they're having within our software products, right?
• 39:19
  SPEAKER_01: Um, so again, just sort of foreshadowing what's what's coming in the future, it's probably not gonna come as a surprise to anybody who's following along because it's you know it's it's happening in other industries as well, right?
• 39:30
  SPEAKER_01: But um, you know, there's there's a obviously tremendous, uh tremendous use case for that.
• 39:36
  SPEAKER_00: Um Yeah, I'm kind of excited to see, you know, because we we're obviously talking and actively promoting agency copilots from a kind of front-end sales perspective, from a back office kind of application perspective.
• 39:48
  SPEAKER_00: But as you know, one of the things we do really well is is the whole kind of configuration piece.
• 39:52
  SPEAKER_00: And I get the question a lot come up that wow, that that does seem I can see that it's no code and I can see that we can get there, but it looks a little complicated.
• 40:00
  SPEAKER_00: We know that after a couple of weeks and of of training, we've had you know customers that have we've helped them build the first two and they've gone on to build another 30 plus on their own.
• 40:08
  SPEAKER_00: But then you know, you're right, to have that kind of that kind of AI capability where we could potentially be feeding it and teaching them and and them actually kind of you know feeding in their experiences into it and teaching themselves, because actually there's plenty of um clients where we kind of log in and look and go, oh wow, didn't look what they've done there.
• 40:25
  SPEAKER_00: That that's cool.
• 40:26
  SPEAKER_00: We'd never thought of that.
• 40:27
  SPEAKER_00: You know, it's you know, because let's face it, some of this config configuration capability is is kind of endless as to as to where you can uh you can take it and what you can do with it.
• 40:35
  unknown: Yeah.
• 40:36
  SPEAKER_00: So yeah, that's exciting.
• 40:38
  SPEAKER_01: And a double-edged sword there, right?
• 40:40
  SPEAKER_01: Because again, I mean, you know, you can you can over-engineer things, you can over-generate things, right?
• 40:45
  SPEAKER_01: So um again, that's where the process uh controls need to need to be you know properly thought through and implemented.
• 40:55
  SPEAKER_01: Um, you know, it just it just it just shifts the the burden from from one category to the next, right?
• 41:02
  SPEAKER_01: Definitely, definitely.
• 41:03
  SPEAKER_01: Yeah.
• 41:04
  SPEAKER_01: Good.
• 41:04
  SPEAKER_01: Would uh yeah, any any closing thoughts?
• 41:07
  SPEAKER_00: I've got a question.
• 41:08
  SPEAKER_00: Um it might turn into a final kind of discussion point, and it came up.
• 41:13
  SPEAKER_00: I've seen it on the news today.
• 41:14
  SPEAKER_00: You must have seen it.
• 41:16
  SPEAKER_00: Um what do you think of the uh of the flying car?
• 41:19
  SPEAKER_00: Um that Mr.
• 41:21
  SPEAKER_00: Musk has uh released to the world that can do flips, drive upside down, and jump over lorries.
• 41:28
  SPEAKER_01: Yeah.
• 41:28
  SPEAKER_00: Who's who's insuring that one?
• 41:31
  SPEAKER_01: Yeah, yeah.
• 41:32
  SPEAKER_01: Yeah, I guess within what context.
• 41:33
  SPEAKER_01: I mean, what do I think about it within what context, right?
• 41:36
  SPEAKER_01: Um you know, I think I think I think um the world is changing rapidly, and I think that this is gonna be a really, really boring answer.
• 41:51
  SPEAKER_01: Um I think that regulators are much more attuned to uh needing to react quickly to these things.
• 42:01
  SPEAKER_01: I I think the the regulatory environment has can probably be you know accused of being slow historically, and I think that they are uh you know much more reactive um to recent changes, so uh or the speed of change at which things are happening in society.
• 42:23
  SPEAKER_01: So, you know, who's gonna insure a flying car?
• 42:26
  SPEAKER_01: What category of insurance does it fall within?
• 42:29
  SPEAKER_01: Um it doesn't exist, but it's going to have to exist uh because the time to market these things is is so much more quick now.
• 42:38
  SPEAKER_01: Um and so the regulatory environment's gonna react accordingly.
• 42:43
  SPEAKER_00: That's so true.
• 42:44
  SPEAKER_00: And and you you know the point you made there was the biggest thing I thought of, which is is it motor insurance, is it aviation insurance?
• 42:50
  SPEAKER_00: God knows.
• 42:50
  SPEAKER_00: But you're right.
• 42:51
  SPEAKER_00: Insurance has always been able to adapt to change.
• 42:53
  SPEAKER_00: It's one of the things it does better than probably most industries.
• 42:56
  SPEAKER_00: So I'm sure we'll uh we'll get plenty of updates as to uh as to how that progresses alongside uh Elon's new project.
• 43:02
  SPEAKER_00: But I mean I did look at it today, my son came running in the room and said, Oh, look at this, dad.
• 43:06
  SPEAKER_00: You know, it sticks to the road and it does this.
• 43:08
  SPEAKER_00: I'm like, yeah, but why?
• 43:10
  SPEAKER_00: If it can stay like that so I can get around the M25, um, you know, then I'm I'm interested.
• 43:16
  SPEAKER_00: Then I'm gonna buy it.
• 43:17
  SPEAKER_00: Jumping over one or two cars, yeah, I don't think so.
• 43:20
  SPEAKER_01: Yeah, I I mean I think I think uh if it does that, everybody's gonna jump over the M25 and you're just gonna have a brand new set of traffic constraints, right?
• 43:28
  SPEAKER_00: So uh well you you'd you'd just stick to the old car and go underneath them, won't you?
• 43:32
  SPEAKER_01: Yeah, I'm a I'm uh I'm a bit of I'm a bit of a purist where where these things are concerned.
• 43:37
  SPEAKER_01: So um I'm happy to hear the sound of of an engine.
• 43:42
  SPEAKER_01: Uh you know, uh and and I'm happy to keep my four wheels on on the on the the pavement.
• 43:49
  SPEAKER_01: Um yeah.
• 43:50
  SPEAKER_00: That means two of us.
• 43:52
  SPEAKER_00: That makes two of us.
• 43:53
  SPEAKER_01: Yeah.
• 43:53
  SPEAKER_01: Well, good, good, good close, uh Lloyd.
• 43:56
  SPEAKER_01: Uh that was fun.
• 43:58
  SPEAKER_01: Hopefully, hopefully by the time we get to next week's episode, uh, you know, we're not actually having to debate uh you know these things being in production.
• 44:07
  SPEAKER_01: So cool.
• 44:10
  SPEAKER_01: Cool.
• 44:11
  SPEAKER_01: All right, but listen, Lloyd.
• 44:12
  SPEAKER_01: Have a great weekend.
• 44:13
  SPEAKER_01: Uh and uh yeah, until next week, everyone.
• 44:15
  SPEAKER_01: Thanks for tuning in.
• 44:16
  SPEAKER_01: Thanks, Elliot.
• 44:17
  SPEAKER_01: Enjoy the marathon.
• 44:18
  SPEAKER_01: Take care.
• 44:18
  SPEAKER_01: Yeah, cheers.
• 44:19
  SPEAKER_01: Bye.
• 44:20
  SPEAKER_01: Bye buddy.