"AI isn't magic": Making it Work for Insurance with Charles Taylor InsureTech

The insurance industry is drowning in data, and manual processes just aren't cutting it anymore. In this episode, we sit down with Reiss Gavin from Charles Taylor InsureTech to uncover how they're helping their customers break free from legacy systems to drive real innovation.

We dive deep into the creation of Bordereaux Sync, a game-changing solution built with the Transparity AI Factory that automates the headache of data cleansing and validation. Reiss shares the challenges facing the insurance industry and his insights on the road ahead, including the AI-generated challenges coming next.

In this episode:

• From chaos to clarity: How to operationalise data cleansing and governance without the months-long headache.
• Real-world deployment: Leveraging AI to solve persistent bottlenecks in claims automation, underwriting accuracy, and fraud detection.
• Architecture over hype: Why successful AI integration depends on strategic collaboration and domain expertise, not just throwing more tech at the problem.

The Transparity Transform Podcast brings together thought leaders from across the UK's technology sector to discuss the industry and nation's hottest topics.

Brought to you by Transparity, the UK's most accredited pureplay Microsoft partner. Find out more about who we are and what we do, at www.transparity.com 

Transcript

1

00:00:01.965 --> 00:00:03.895

Welcome to the Transform podcast.

2

00:00:04.365 --> 00:00:07.255

Delighted to have Reese Gavin here from Charles Taylor,

3

00:00:07.635 --> 00:00:10.375

and we'll talk through some of what's going on in InsureTech

4

00:00:10.595 --> 00:00:12.495

as well as a bit on his experience

5

00:00:12.495 --> 00:00:14.255

with Transparity's AI Factory.

6

00:00:14.875 --> 00:00:18.295

I'm the head of London Market application development

7

00:00:18.675 --> 00:00:20.895

design for Charles Taylor InsureTech.

8

00:00:21.395 --> 00:00:24.415

So amazing to have you here, Reese. Thank you. Thank you.

9

00:00:24.655 --> 00:00:26.895

I think we are really excited just to kind

10

00:00:26.895 --> 00:00:29.895

of dig into the context of where you guys are coming from,

11

00:00:30.285 --> 00:00:31.895

from the InsureTech space.

12

00:00:32.435 --> 00:00:34.495

Um, so wanted to start with just some of the challenges

13

00:00:34.495 --> 00:00:37.135

that you guys are facing, doing more with less,

14

00:00:37.615 --> 00:00:40.575

a bit on needing to basically, you know, facing pressure

15

00:00:40.575 --> 00:00:42.855

to modernize legacy applications.

16

00:00:43.435 --> 00:00:45.175

Um, where are you finding your,

17

00:00:45.285 --> 00:00:47.975

your organization at this point in time?

18

00:00:48.515 --> 00:00:51.015

Um, well, I think organizationally it's quite

19

00:00:51.015 --> 00:00:52.215

an exciting time at the moment.

20

00:00:52.215 --> 00:00:54.455

There's a lot happening in the technology space.

21

00:00:54.455 --> 00:00:58.575

Technology is adapting and advancing at a a rapid pace.

22

00:00:58.915 --> 00:01:03.575

Um, and, you know, we primarily work in the London market

23

00:01:03.715 --> 00:01:08.455

and the uk but we have clients across Latin America, uh,

24

00:01:08.995 --> 00:01:12.055

the US who, who have different challenges.

25

00:01:12.515 --> 00:01:16.975

So obviously our core is, um, technology products

26

00:01:16.995 --> 00:01:18.335

for the insurance industry.

27

00:01:18.875 --> 00:01:21.935

And that can be from policy administration systems,

28

00:01:22.465 --> 00:01:24.415

fraud systems, uh,

29

00:01:24.415 --> 00:01:26.855

delegated authorities specifically for the London market.

30

00:01:27.835 --> 00:01:30.335

Um, and obviously now venturing into AI

31

00:01:30.395 --> 00:01:33.695

and how AI can enhance our systems

32

00:01:34.155 --> 00:01:36.015

and add value to our clients.

33

00:01:36.875 --> 00:01:41.515

Um, but also keeping in mind that AI is not magic. Yeah.

34

00:01:42.155 --> 00:01:46.475

And, um, while there's a, a big drive to embrace it

35

00:01:46.475 --> 00:01:48.755

and see how it can be used within very targeted

36

00:01:48.815 --> 00:01:52.875

and using it in ways that enhance organization's ability,

37

00:01:53.375 --> 00:01:57.315

uh, adds efficiency to them, uh, without building a solution

38

00:01:57.315 --> 00:01:58.555

and then going looking for a problem.

39

00:01:59.635 --> 00:02:01.555

Absolutely. I think that makes a lot of sense.

40

00:02:01.815 --> 00:02:04.755

Um, you know, pressure to do more with less is something

41

00:02:04.895 --> 00:02:07.155

that's existed in insurance for a long time.

42

00:02:07.335 --> 00:02:10.155

Oh, yes. And it's finding out like the right solutions at

43

00:02:10.155 --> 00:02:12.875

the right time and working with you guys has been a bit

44

00:02:12.875 --> 00:02:14.675

of a passion project for me personally.

45

00:02:15.295 --> 00:02:18.995

Um, my dad worked his entire career in reinsurance. Oh.

46

00:02:19.135 --> 00:02:22.595

And he actually developed, um, a product that they,

47

00:02:22.665 --> 00:02:25.355

they still use a piece of technology that's, you know,

48

00:02:25.355 --> 00:02:26.715

was he developed in the nineties

49

00:02:26.775 --> 00:02:30.315

and is still being used to, you know, rectify different, um,

50

00:02:30.495 --> 00:02:32.555

things from a financial standpoint. So

51

00:02:32.815 --> 00:02:34.795

Sounds like a lot of insurance systems from

52

00:02:34.795 --> 00:02:35.875

the nineties still being used.

53

00:02:36.185 --> 00:02:38.435

Yeah. And that's where the opportunity, I think

54

00:02:38.435 --> 00:02:39.795

to work together came up.

55

00:02:40.415 --> 00:02:44.075

Um, where did transparency kind of come in to this world

56

00:02:44.255 --> 00:02:47.755

of looking to leverage ai but in the right places? Okay,

57

00:02:48.375 --> 00:02:51.555

So obviously, um, our main platform we,

58

00:02:51.555 --> 00:02:54.235

based on Microsoft, so we have quite a good relationship

59

00:02:54.235 --> 00:02:56.755

with Microsoft, and we were looking in the AI space,

60

00:02:56.815 --> 00:02:59.325

and we obviously have our own developers in that,

61

00:02:59.425 --> 00:03:00.565

but it's emerging, and

62

00:03:01.145 --> 00:03:03.365

that's not necessarily at the moment their core skillset

63

00:03:03.585 --> 00:03:07.565

around, um, AI and, uh, AgTech and agents

64

00:03:07.705 --> 00:03:08.805

and Python and that.

65

00:03:09.425 --> 00:03:11.525

So we had lots of discussions with Microsoft

66

00:03:11.745 --> 00:03:14.325

and they obviously, as you know, as you're one,

67

00:03:14.325 --> 00:03:15.725

have their technology partners.

68

00:03:16.425 --> 00:03:19.845

Um, so they recommended three to us.

69

00:03:20.465 --> 00:03:23.565

Um, and we had a bit of a, uh, so we say dog

70

00:03:23.565 --> 00:03:26.685

and pony show where people came and put their wares out there.

71

00:03:26.685 --> 00:03:29.485

And we, we gave them the concept and the idea

72

00:03:30.345 --> 00:03:33.005

and basically let the three candidates come back

73

00:03:33.385 --> 00:03:35.565

and to be fair, Transparity were head

74

00:03:35.565 --> 00:03:36.925

and shoulders above the rest.

75

00:03:37.105 --> 00:03:40.045

And not only their understanding of what we were trying

76

00:03:40.045 --> 00:03:43.125

to achieve, but also the model they presented

77

00:03:43.125 --> 00:03:46.305

and how they would take, um, the concept

78

00:03:46.885 --> 00:03:48.105

to a proof of value.

79

00:03:48.845 --> 00:03:51.585

Um, so really you guys won it on your own merits

80

00:03:51.585 --> 00:03:52.985

and the way the engagement and,

81

00:03:52.985 --> 00:03:54.625

and coming back and the understanding.

82

00:03:55.205 --> 00:03:56.865

And I think our decision to go

83

00:03:56.865 --> 00:04:00.225

with Transparity was born out where we had a number of

84

00:04:00.945 --> 00:04:03.425

workshops at, um, upstairs at Microsoft

85

00:04:03.605 --> 00:04:05.585

and, um, obviously in our offices,

86

00:04:05.965 --> 00:04:09.465

and we, we came up with, if you like, a very

87

00:04:12.355 --> 00:04:15.465

simplistic straw man after a couple of workshops.

88

00:04:15.565 --> 00:04:18.025

And, uh, obviously we had some conversations and,

89

00:04:18.125 --> 00:04:19.305

and transparency came back

90

00:04:19.305 --> 00:04:20.465

and said, yeah, we'll have a proof

91

00:04:20.465 --> 00:04:22.305

of value working in 30 days.

92

00:04:22.875 --> 00:04:26.105

Which having worked in technology for about 20 years,

93

00:04:26.305 --> 00:04:28.065

I was really skeptical that, to be fair,

94

00:04:28.725 --> 00:04:29.945

but true to your words, you did

95

00:04:30.445 --> 00:04:34.105

and you actually exceeded expectations that did a lot more

96

00:04:34.725 --> 00:04:38.505

in a functionally rich way than we were expecting for,

97

00:04:38.645 --> 00:04:39.865

if you want, as a prototype,

98

00:04:39.865 --> 00:04:41.025

which I know everyone hates the word,

99

00:04:41.045 --> 00:04:42.305

but a, a proof of value.

100

00:04:43.085 --> 00:04:45.025

Um, so that sort of reinforced

101

00:04:45.025 --> 00:04:46.545

and cemented that we'd made the right decision.

102

00:04:46.765 --> 00:04:48.385

And then as we got further into it,

103

00:04:48.385 --> 00:04:51.745

which I'm sure will come onto around going from a, a proof

104

00:04:51.745 --> 00:04:55.905

of value to a, a working product, um, you know,

105

00:04:56.375 --> 00:04:57.865

it's the model that you employ

106

00:04:57.865 --> 00:05:00.025

and the way you work is, is somewhat refreshing,

107

00:05:00.025 --> 00:05:01.385

but it actually delivers results.

108

00:05:02.055 --> 00:05:03.515

That's really fantastic. Yeah.

109

00:05:03.835 --> 00:05:05.835

I remember when, uh, we kind of were working through

110

00:05:05.835 --> 00:05:06.835

that process with you

111

00:05:06.835 --> 00:05:08.675

and you're like, well, if you can do

112

00:05:08.675 --> 00:05:11.435

what you say you're able to do, that'll be fantastic.

113

00:05:11.785 --> 00:05:14.195

Yeah. And then, yeah, proving that out was really important

114

00:05:14.195 --> 00:05:16.035

to us, so that's fantastic to hear.

115

00:05:16.635 --> 00:05:19.515

I think let's delve into the actual solution a little bit.

116

00:05:19.515 --> 00:05:22.475

Mm-hmm. So for people that aren't as familiar

117

00:05:22.545 --> 00:05:23.955

with the insurance process

118

00:05:24.095 --> 00:05:28.155

and four to row, we've kind of mentioned, um, what, what was

119

00:05:28.315 --> 00:05:30.635

that challenge that you were seeing from your customer base?

120

00:05:30.905 --> 00:05:33.275

Okay, so if I take a step back for people who,

121

00:05:33.335 --> 00:05:34.835

who may not be aware board, you know,

122

00:05:35.635 --> 00:05:39.195

Bordeaux is a very easy way for

123

00:05:40.385 --> 00:05:42.035

brokers or cover holders

124

00:05:42.495 --> 00:05:46.715

or TPAs dcas to give a lot of data

125

00:05:46.895 --> 00:05:49.675

to the, the person who's underwriting the insurance.

126

00:05:50.105 --> 00:05:52.635

It's just an Excel spreadsheet with lots of columns on it

127

00:05:52.635 --> 00:05:53.795

and lots of rows of data.

128

00:05:54.655 --> 00:05:56.755

Um, now that's really good

129

00:05:57.225 --> 00:05:59.285

for moving vast volumes of data easily.

130

00:06:00.115 --> 00:06:01.765

It's really bad for data quality

131

00:06:01.875 --> 00:06:03.125

because at the end of the day,

132

00:06:03.125 --> 00:06:04.525

you can put anything you want on an Excel

133

00:06:04.525 --> 00:06:06.165

spreadsheet, doesn't matter.

134

00:06:07.025 --> 00:06:09.325

Um, now we have a Bordereaux management solution,

135

00:06:09.325 --> 00:06:13.445

which takes those spreadsheets, it processes them, cleanses

136

00:06:13.445 --> 00:06:14.525

and validates the data,

137

00:06:15.465 --> 00:06:19.125

and then gives it, um, data feeds into a, a data warehouse.

138

00:06:19.125 --> 00:06:21.685

And then insurers can take and, and do that

139

00:06:21.685 --> 00:06:25.485

because Bordereaux exists as insurers write business worldwide,

140

00:06:25.785 --> 00:06:27.285

but they don't necessarily have a license

141

00:06:27.465 --> 00:06:29.445

to write the business in all the jurisdictions.

142

00:06:29.705 --> 00:06:31.605

So they have what's called delegated authority,

143

00:06:31.895 --> 00:06:34.565

which is effectively going to someone, you,

144

00:06:34.625 --> 00:06:36.525

you have a license to operate in that territory.

145

00:06:36.745 --> 00:06:38.525

You can sell insurance on my behalf.

146

00:06:38.865 --> 00:06:40.765

That's simplistically what it is. Mm-hmm.

147

00:06:41.465 --> 00:06:44.925

Um, so our Bordereaux management solution,

148

00:06:44.925 --> 00:06:46.805

and I'm sure a lot of them out there when they process

149

00:06:46.875 --> 00:06:49.285

that data, they will find errors, they will find

150

00:06:50.005 --> 00:06:51.085

failings in the rules.

151

00:06:51.315 --> 00:06:53.285

They will find breaches and limits.

152

00:06:53.835 --> 00:06:55.405

Some of the data can be fixed in the systems,

153

00:06:55.785 --> 00:06:59.125

but a lot of the times you have to back the data out

154

00:06:59.125 --> 00:07:01.685

of the system and you have to go back to the person

155

00:07:01.745 --> 00:07:03.325

who sent it to you to get the corrections.

156

00:07:03.665 --> 00:07:07.285

So it's inefficient. It takes a lot of manual effort.

157

00:07:08.185 --> 00:07:12.525

Um, and unfortunately a lot of systems will allow users

158

00:07:12.625 --> 00:07:15.565

to accept bad data so

159

00:07:15.565 --> 00:07:18.605

that then bad data then flows into downstream systems.

160

00:07:19.025 --> 00:07:22.005

And over time you can find firms making

161

00:07:22.285 --> 00:07:23.565

decisions on dubious data.

162

00:07:23.745 --> 00:07:26.165

Mm-hmm. Um, and I've got lots of examples of that.

163

00:07:26.945 --> 00:07:31.565

Um, so this is a real problem for our customers

164

00:07:31.585 --> 00:07:33.725

and it's a real problem for the insurance industry.

165

00:07:34.145 --> 00:07:36.965

Um, so we looked at it

166

00:07:37.065 --> 00:07:40.605

and thought, how can we solve this problem

167

00:07:41.025 --> 00:07:42.645

or try and solve this problem?

168

00:07:43.945 --> 00:07:47.445

And AI was sort of becoming slightly more embedded.

169

00:07:47.445 --> 00:07:51.645

It was a slightly more reliable than the earliest LLMs.

170

00:07:52.385 --> 00:07:55.725

Um, and a lot of the data issues are either

171

00:07:56.475 --> 00:08:00.565

someone's transposed dates, they've transposed numbers,

172

00:08:00.835 --> 00:08:02.885

they've put data in the wrong columns,

173

00:08:03.625 --> 00:08:05.365

and there's a wealth of information out there

174

00:08:05.365 --> 00:08:06.645

that AI can use.

175

00:08:07.485 --> 00:08:10.625

Um, so we created an application called Bordereaux Sync,

176

00:08:11.035 --> 00:08:13.025

which is AI powered

177

00:08:13.525 --> 00:08:17.745

and uses AI to correct the data as best it can.

178

00:08:18.005 --> 00:08:20.705

Now, I'm not suggesting someone just trusts AI.

179

00:08:21.405 --> 00:08:24.225

So part of the application will also tell the user

180

00:08:24.225 --> 00:08:26.625

what it's corrected, what the original value was,

181

00:08:27.125 --> 00:08:29.825

and it produces an AI created board.

182

00:08:30.645 --> 00:08:33.145

Um, the user can then process the AI board

183

00:08:33.805 --> 00:08:35.745

or process the original wardrobe,

184

00:08:35.765 --> 00:08:38.385

but know they know what all the problems are

185

00:08:39.605 --> 00:08:42.785

and working with transport, it does what it says on the tin.

186

00:08:42.805 --> 00:08:44.265

It does fix the data.

187

00:08:44.565 --> 00:08:46.025

Um, it's border agnostic,

188

00:08:46.025 --> 00:08:48.745

so risk premium claims, risk and premium.

189

00:08:49.605 --> 00:08:51.825

And the brilliance about AI is,

190

00:08:51.925 --> 00:08:54.665

the way we've built it is it only works

191

00:08:54.685 --> 00:08:55.725

on a customer's data.

192

00:08:55.945 --> 00:08:57.725

We don't share multi customer data.

193

00:08:58.265 --> 00:09:00.405

So it's always learning on a customer's data.

194

00:09:00.825 --> 00:09:05.805

And because ai, AI can learn, it's not reliant on a human

195

00:09:05.825 --> 00:09:08.365

to keep telling it, it's tell it the answer where a lot

196

00:09:08.365 --> 00:09:11.045

of the Bordereaux management solutions need a human

197

00:09:11.045 --> 00:09:13.925

to tell it the answer, but it still has that check

198

00:09:13.925 --> 00:09:15.605

and balance where it will go back to a human

199

00:09:15.625 --> 00:09:17.885

and say, Hey, I found this and I've changed it to this.

200

00:09:18.505 --> 00:09:21.245

And the other beauty of it is it has the ability,

201

00:09:21.505 --> 00:09:25.525

if it knows who to contact, can go back to the person

202

00:09:25.545 --> 00:09:27.965

who submitted the data, tell them what the error is,

203

00:09:28.185 --> 00:09:31.065

ask them for the answer, well, the correction,

204

00:09:31.245 --> 00:09:33.305

and then input that correction.

205

00:09:34.155 --> 00:09:36.375

So that's taking, you know,

206

00:09:37.335 --> 00:09:40.655

a very laborious manual task away from someone

207

00:09:40.655 --> 00:09:41.855

who would otherwise have to do it.

208

00:09:42.735 --> 00:09:44.975

Absolutely. Yeah. I think this is such a great example

209

00:09:45.235 --> 00:09:48.175

for people listening to think about different processes

210

00:09:48.205 --> 00:09:51.455

that might be onerous, might have multiple people engaged.

211

00:09:52.075 --> 00:09:55.415

Um, another element of this is kind of the data element.

212

00:09:55.635 --> 00:09:56.855

So keeping data clean

213

00:09:57.155 --> 00:09:58.455

and making sure that decisions

214

00:09:58.455 --> 00:10:00.055

that are follow on decisions are made

215

00:10:00.055 --> 00:10:01.175

off the right information.

216

00:10:01.205 --> 00:10:03.935

Yeah, indeed. So those are a lot, those are kind of through,

217

00:10:04.085 --> 00:10:08.255

through, um, lines that I think are really helpful and,

218

00:10:08.275 --> 00:10:09.895

and kind of thinking about okay,

219

00:10:09.895 --> 00:10:13.135

what might be a good application of AI to, you know,

220

00:10:13.255 --> 00:10:14.975

a project or a problem that we're kind

221

00:10:14.975 --> 00:10:16.215

of working with. Yeah.

222

00:10:16.475 --> 00:10:18.695

And, and I guess the beauty of our application,

223

00:10:18.695 --> 00:10:22.455

while it's primarily being built and designed for boro

224

00:10:22.455 --> 00:10:24.415

and delegated authority, at the end

225

00:10:24.415 --> 00:10:28.735

of the day it's validating data so we can quite easily pivot

226

00:10:29.555 --> 00:10:33.235

and use that, um, functionality

227

00:10:33.455 --> 00:10:36.875

and technology other over other data sets.

228

00:10:37.015 --> 00:10:38.555

And we've obviously got a number

229

00:10:38.615 --> 00:10:41.995

of different towers within Charles Taylor or CTI.

230

00:10:42.375 --> 00:10:45.355

Um, so I'm having conversations with my colleagues who,

231

00:10:45.495 --> 00:10:49.075

you know, one of the, one of the teams that put our, uh,

232

00:10:49.075 --> 00:10:50.875

policy admin system in and develop

233

00:10:50.895 --> 00:10:52.715

and deploy that, you know,

234

00:10:52.715 --> 00:10:54.875

they're finding their customers have similar sort

235

00:10:54.875 --> 00:10:55.915

of challenges around data.

236

00:10:55.975 --> 00:10:59.395

So we're looking at seeing, well, how, how can our learnings

237

00:10:59.395 --> 00:11:04.155

and our technology be pivoted to support other towers

238

00:11:04.155 --> 00:11:07.435

and their capabilities and needs of their clients?

239

00:11:08.475 --> 00:11:10.275

Absolutely. Yeah. That's really, really cool.

240

00:11:10.275 --> 00:11:12.835

Like taking the learnings, I mean, similar to

241

00:11:12.835 --> 00:11:15.475

what the actual Bordereaux Sync does is you're taking these

242

00:11:15.635 --> 00:11:17.915

learnings and then you can apply them to other elements,

243

00:11:18.175 --> 00:11:19.275

but with safety and,

244

00:11:19.275 --> 00:11:22.635

and within kind of the, the plans that you guys have, um,

245

00:11:22.815 --> 00:11:25.555

to serve your clients across, across the firm.

246

00:11:26.215 --> 00:11:29.875

Um, digging a little bit into making the Bordeaux Sync

247

00:11:29.945 --> 00:11:31.235

product more real.

248

00:11:31.585 --> 00:11:34.435

Obviously you engage Transparity in our AI factory.

249

00:11:35.055 --> 00:11:37.435

Um, and just wanted to know a little bit more about your

250

00:11:37.435 --> 00:11:39.235

experience with the AI factory process.

251

00:11:39.825 --> 00:11:42.155

What differentiated it with maybe other ways

252

00:11:42.395 --> 00:11:44.435

that you've worked with other partners in the past?

253

00:11:44.935 --> 00:11:46.075

Um, yeah, what's it been like?

254

00:11:46.335 --> 00:11:48.355

Uh, it's been really good, really positive.

255

00:11:48.775 --> 00:11:51.675

Um, and obviously my team is working closely

256

00:11:51.745 --> 00:11:53.925

with the AI factory team at Transparity.

257

00:11:54.185 --> 00:11:57.045

But, you know, three key things that stand out to me

258

00:11:57.345 --> 00:12:00.445

for it is, one, the flexibility to change, um,

259

00:12:00.745 --> 00:12:03.085

the collaborative development, which is really important

260

00:12:03.385 --> 00:12:07.605

and the complexity complex challenges into workable designs.

261

00:12:07.625 --> 00:12:10.005

And, and to touch on those, um,

262

00:12:10.795 --> 00:12:12.845

because the tool is new

263

00:12:13.345 --> 00:12:16.485

and we are having, um, we sort of test and learn

264

00:12:16.485 --> 00:12:18.965

and we're having sort of trialing it with some

265

00:12:18.965 --> 00:12:20.925

of our customers, getting their feedback.

266

00:12:21.835 --> 00:12:25.325

It's, it's a moving target almost.

267

00:12:26.105 --> 00:12:30.325

Um, and what we found with the AI factory so far,

268

00:12:30.335 --> 00:12:33.845

touch wood, is that no matter the, the change

269

00:12:33.845 --> 00:12:36.925

that we throw at them, they, the way they operate,

270

00:12:36.995 --> 00:12:39.845

they're very, um, flexible

271

00:12:39.945 --> 00:12:42.125

and being able to pivot and making changes.

272

00:12:42.905 --> 00:12:45.925

So much. So we had one example where we did a demo

273

00:12:46.065 --> 00:12:47.165

to a client in the morning

274

00:12:47.585 --> 00:12:50.645

and we sort of spotted something and we spoke to them.

275

00:12:50.645 --> 00:12:52.805

We had another demo to the client in the afternoon,

276

00:12:52.825 --> 00:12:55.005

and it was a bit odd just going, oh, look, they've fixed it.

277

00:12:56.065 --> 00:12:58.325

You know, so it is that real time, instantaneous,

278

00:12:58.395 --> 00:12:59.885

instantaneous change.

279

00:13:00.025 --> 00:13:02.085

And I think that comes from having a dedicated team at the

280

00:13:02.085 --> 00:13:04.125

transparity side, a dedicated team in the

281

00:13:04.265 --> 00:13:05.605

the CTI side as well.

282

00:13:05.665 --> 00:13:08.165

And, you know, they work together in terms

283

00:13:08.165 --> 00:13:09.405

of collaborative development.

284

00:13:09.705 --> 00:13:12.245

Um, the big thing we found is we,

285

00:13:12.385 --> 00:13:14.245

we haven't had death by workshop.

286

00:13:14.675 --> 00:13:17.565

It's true. Always been like there's been one very intensive

287

00:13:17.785 --> 00:13:18.885

day, two day workshop.

288

00:13:19.105 --> 00:13:20.125

And you know, to be fair,

289

00:13:20.245 --> 00:13:22.405

I don't think transparity our insurance experts

290

00:13:22.905 --> 00:13:25.365

by any sense of the imagination,

291

00:13:26.065 --> 00:13:27.845

but they took time to understand the problem

292

00:13:27.945 --> 00:13:30.605

and that early understanding the concepts and the problem,

293

00:13:30.745 --> 00:13:31.845

and not being scared

294

00:13:31.945 --> 00:13:33.885

to say when they don't understand something

295

00:13:34.065 --> 00:13:35.285

or ask for clarification.

296

00:13:36.185 --> 00:13:41.115

Um, and the, the beauty of that is,

297

00:13:41.115 --> 00:13:42.715

is because we have a good working relationship

298

00:13:42.715 --> 00:13:46.235

and is dedicated, it's not a

299

00:13:47.275 --> 00:13:49.955

customer vendor relationship in the traditional sense.

300

00:13:50.015 --> 00:13:51.315

It, it's more one team.

301

00:13:52.135 --> 00:13:54.715

Um, and the way that we, you know,

302

00:13:55.015 --> 00:13:58.355

not only are we leveraging transparity to the early days

303

00:13:58.355 --> 00:14:00.355

of development, the progression and the robust,

304

00:14:00.495 --> 00:14:01.795

but there's also a substream,

305

00:14:01.795 --> 00:14:03.355

which is the knowledge transfer

306

00:14:03.695 --> 00:14:07.475

and being able to take, you know, more ownership ourselves.

307

00:14:07.775 --> 00:14:10.155

Um, but unfortunately the speed of development

308

00:14:10.155 --> 00:14:12.435

that's take you a little bit longer than desire,

309

00:14:12.495 --> 00:14:13.515

but you know, that's life.

310

00:14:13.905 --> 00:14:16.835

Yeah. That's a big piece of the AI factory as well

311

00:14:16.985 --> 00:14:19.235

that we find is that kind of knowledge transfer,

312

00:14:19.425 --> 00:14:22.315

helping our clients understand what tech to use

313

00:14:22.315 --> 00:14:24.875

and when, why we've made the decisions that we've had.

314

00:14:24.895 --> 00:14:26.395

Mm-hmm. So I think that's kind

315

00:14:26.395 --> 00:14:27.635

of a, a really helpful element.

316

00:14:27.775 --> 00:14:29.955

And yeah, I would say that it's felt like one team

317

00:14:30.065 --> 00:14:31.315

from our standpoint as well.

318

00:14:31.315 --> 00:14:33.795

Yeah. One thing I thought that you did that was brilliant

319

00:14:33.795 --> 00:14:36.435

that we were just talking about was, um, the fact

320

00:14:36.435 --> 00:14:38.275

that you brought in it,

321

00:14:38.895 --> 00:14:41.275

but you also brought in the sales guys Yeah.

322

00:14:41.275 --> 00:14:42.995

That were going to be selling the solution

323

00:14:42.995 --> 00:14:46.195

and pitching it, who really understand the client problems

324

00:14:46.265 --> 00:14:47.915

into at least one of our workshops.

325

00:14:48.175 --> 00:14:50.675

And then they've been kind of off to the races in terms

326

00:14:50.675 --> 00:14:52.165

of presenting the right things

327

00:14:52.165 --> 00:14:53.605

to the right clients at the right time.

328

00:14:53.835 --> 00:14:56.085

Yeah. Um, so I wanted to learn a little bit more about

329

00:14:56.085 --> 00:14:58.525

how you orchestrated that from your point of view,

330

00:14:58.525 --> 00:15:00.405

because a big thing that we talk about

331

00:15:00.435 --> 00:15:03.925

with these workshops is really getting different points

332

00:15:03.925 --> 00:15:05.405

of view into the room together.

333

00:15:05.745 --> 00:15:07.245

So how did you pull that off? Well,

334

00:15:08.485 --> 00:15:10.085

I say, I just told them, okay,

335

00:15:10.085 --> 00:15:11.085

Yeah, yeah. Volunteering

336

00:15:11.085 --> 00:15:12.005

works. Yeah.

337

00:15:12.465 --> 00:15:14.525

You volunteered. No, uh, no.

338

00:15:14.525 --> 00:15:17.125

I look at, in organizations, there's many functions

339

00:15:17.125 --> 00:15:19.085

and each function has experts in it.

340

00:15:19.265 --> 00:15:22.285

Um, but I'm a great believer in, you know, no silos.

341

00:15:22.705 --> 00:15:25.485

And even if someone sits in a two hour workshop

342

00:15:25.485 --> 00:15:28.845

or a two day workshop that can be technical,

343

00:15:29.035 --> 00:15:30.365

they will glean something.

344

00:15:30.865 --> 00:15:32.965

Um, and it could be just that little nugget

345

00:15:33.115 --> 00:15:35.365

that is the difference between a sale and not a sale.

346

00:15:35.985 --> 00:15:37.245

And also, you know,

347

00:15:37.245 --> 00:15:38.925

you can have the world's greatest product,

348

00:15:39.985 --> 00:15:42.285

but if you can't market it and you can't sell it,

349

00:15:42.285 --> 00:15:44.565

and the people selling it don't understand it,

350

00:15:45.355 --> 00:15:46.445

it's gonna go nowhere.

351

00:15:47.265 --> 00:15:50.285

So it's good to get that feedback from say,

352

00:15:50.285 --> 00:15:52.725

the non-technical people around, you know, give us,

353

00:15:52.875 --> 00:15:55.965

give us a, a scenario.

354

00:15:56.185 --> 00:15:57.245

How would you go and sell this?

355

00:15:57.245 --> 00:15:59.045

What, what, what do you need to show?

356

00:15:59.275 --> 00:16:01.525

Because you get from the, the designers

357

00:16:01.525 --> 00:16:03.565

and the developers, they might know that intimately

358

00:16:03.565 --> 00:16:06.885

and how it works, but they've got no idea of how to sell it.

359

00:16:07.185 --> 00:16:08.885

So again, you know, you have great tech,

360

00:16:09.465 --> 00:16:13.405

but no one can visualize it or understand it.

361

00:16:13.425 --> 00:16:15.485

And it's almost like the, the touch feel.

362

00:16:16.065 --> 00:16:20.205

So by getting the, um, the sales guys in there, you know,

363

00:16:20.235 --> 00:16:21.725

they sort of understand it.

364

00:16:21.825 --> 00:16:24.045

And so we, you know, they have good contributions

365

00:16:24.065 --> 00:16:27.285

to make in their own special sales way. Yeah.

366

00:16:27.515 --> 00:16:29.605

They were brilliant. Yeah. I, I totally agree.

367

00:16:29.625 --> 00:16:31.645

And even that felt like one team. Yeah.

368

00:16:31.805 --> 00:16:33.525

'cause sometimes you'll have yeah.

369

00:16:33.625 --> 00:16:36.285

It or sales and bring in kind of cross-functional groups,

370

00:16:36.385 --> 00:16:38.765

but we were all driving towards one thing.

371

00:16:38.955 --> 00:16:42.045

Yeah. Um, and yeah, I, I really remember them. Yeah.

372

00:16:42.565 --> 00:16:43.925

Contributing their own kind of sales way,

373

00:16:43.985 --> 00:16:46.445

but really helpful in thinking about, um,

374

00:16:46.555 --> 00:16:48.925

just different things to add to the, the product.

375

00:16:49.225 --> 00:16:50.245

So that was fantastic.

376

00:16:50.865 --> 00:16:52.565

Um, you also mentioned kind of the pace

377

00:16:52.625 --> 00:16:54.205

and the speed at, at delivery.

378

00:16:54.355 --> 00:16:56.525

Yeah. Um, how was, how important was that to you?

379

00:16:56.625 --> 00:17:00.205

And in terms of like reporting up, um, on

380

00:17:00.205 --> 00:17:01.445

what you're delivering?

381

00:17:01.755 --> 00:17:04.685

Yeah, unfortunately, no, no. It development's free, right?

382

00:17:04.925 --> 00:17:08.925

Um, so, you know, to be fair, CTI has taken a bit

383

00:17:08.925 --> 00:17:11.005

of a risk in terms of investing in this technology.

384

00:17:11.305 --> 00:17:14.805

Um, but it was important to us, you know, to be able

385

00:17:14.805 --> 00:17:16.885

to have something tangible, not the finished working

386

00:17:16.885 --> 00:17:20.165

product, but just conceptually could show

387

00:17:21.265 --> 00:17:22.995

that it did what we said it would do.

388

00:17:23.215 --> 00:17:27.155

And, you know, when you're going in to talk to the, the COO

389

00:17:27.335 --> 00:17:29.915

or the CTO, you know, they're not the people

390

00:17:29.915 --> 00:17:31.755

who are gonna be using the tool, but they're gonna be the

391

00:17:31.755 --> 00:17:33.035

ones that sign the checks, right.

392

00:17:33.175 --> 00:17:34.395

Or make the recommendation.

393

00:17:34.735 --> 00:17:36.835

So they've got to see value and, and,

394

00:17:36.935 --> 00:17:39.195

and you have to pitch it in a way

395

00:17:39.195 --> 00:17:41.275

that this is the benefit it brings to your organization.

396

00:17:41.275 --> 00:17:43.675

You're not really, on one hand, you're selling them a bit

397

00:17:43.675 --> 00:17:46.915

of software, but you're also selling them an eff, you know,

398

00:17:46.945 --> 00:17:49.115

it's either gotta grow their bottom line,

399

00:17:49.995 --> 00:17:51.885

it's gotta reduce their cost,

400

00:17:51.945 --> 00:17:53.925

or it's gotta drive efficiencies,

401

00:17:54.055 --> 00:17:55.965

which in turn reduce their cost.

402

00:17:56.065 --> 00:17:59.525

Now think is not not gonna go anyone's bottom line,

403

00:18:00.105 --> 00:18:01.445

but it's gonna improve their efficiency

404

00:18:01.445 --> 00:18:02.565

and it's gonna reduce their cost.

405

00:18:03.065 --> 00:18:08.045

Um, and if you can show them that, um, how it will do

406

00:18:08.045 --> 00:18:09.805

that as part of the overall sales pitch,

407

00:18:10.385 --> 00:18:12.165

and they can understand that that's a big win,

408

00:18:12.505 --> 00:18:16.045

but also they're not gonna buy into something that they say

409

00:18:16.045 --> 00:18:17.645

that looks really good, when can we have it?

410

00:18:17.745 --> 00:18:20.645

Oh, a year and a half, right? Yeah.

411

00:18:20.995 --> 00:18:22.445

That sweet spot, isn't it? Yeah. Yeah.

412

00:18:22.585 --> 00:18:26.205

And I think working with transparency, we've managed to get

413

00:18:27.225 --> 00:18:29.125

the way the I AI factory operates,

414

00:18:29.125 --> 00:18:31.285

and it sort of brings you back to the complex challenges

415

00:18:32.145 --> 00:18:33.885

is at the end of the day, this is all about data

416

00:18:34.145 --> 00:18:36.965

and AI's really good with data, you know,

417

00:18:37.025 --> 00:18:38.885

and looking at data, understanding data

418

00:18:39.065 --> 00:18:41.405

and highlighting inconsistencies in data.

419

00:18:41.825 --> 00:18:43.565

So when we're using AI for something

420

00:18:43.565 --> 00:18:46.165

that's really good at today, um,

421

00:18:46.705 --> 00:18:49.965

but two, working with transparency and the, and the one team

422

00:18:50.025 --> 00:18:52.565

and the, the workshops, they were able

423

00:18:52.565 --> 00:18:56.885

to understand complex data problems relatively quickly.

424

00:18:57.505 --> 00:18:59.525

And is, you know, I've worked on other IT projects

425

00:18:59.525 --> 00:19:03.805

where you have sort of the business and IT, and it's siloed

426

00:19:04.225 --> 00:19:05.605

and it's death by workshop

427

00:19:05.675 --> 00:19:10.365

because it can't grasp the, the data nuances

428

00:19:10.425 --> 00:19:12.445

and the problems a lot of the time.

429

00:19:13.345 --> 00:19:16.325

Um, that's not actually black and white.

430

00:19:16.325 --> 00:19:17.485

There are people that cross over,

431

00:19:17.545 --> 00:19:19.885

but as it is a generalization, um,

432

00:19:22.185 --> 00:19:25.965

but you guys have been able to understand the vagaries

433

00:19:25.965 --> 00:19:27.325

and the interpretation and

434

00:19:27.325 --> 00:19:30.085

because at the end of the day, it is really unstructured

435

00:19:30.085 --> 00:19:32.165

data that you're starting with, you know,

436

00:19:32.165 --> 00:19:33.565

it might have the illusion of structure,

437

00:19:33.945 --> 00:19:36.605

but if I go back to, it's just an Excel spreadsheet, right?

438

00:19:36.705 --> 00:19:38.085

You can put anything in it. Yeah.

439

00:19:38.145 --> 00:19:42.605

So that's, that's one thing that we found helped the speed.

440

00:19:43.225 --> 00:19:45.965

And also because everything was collaborative

441

00:19:45.965 --> 00:19:47.685

and the planning was collaborative,

442

00:19:48.325 --> 00:19:52.285

although we keep changing the goalposts, um, you know, we,

443

00:19:52.305 --> 00:19:54.685

we fixed on, you know, we we're using Agile,

444

00:19:54.685 --> 00:19:56.965

we're doing in sprints, we know what the, you know,

445

00:19:57.025 --> 00:20:00.325

the releasable artifact at the end of the sprint has to be.

446

00:20:01.185 --> 00:20:05.605

And, and we've also introduced a, uh,

447

00:20:05.605 --> 00:20:09.085

another third party into the equation who have, um,

448

00:20:10.305 --> 00:20:13.125

and call it AI slash automated testing tool.

449

00:20:13.665 --> 00:20:16.645

And again, we've worked collaboratively with transparency

450

00:20:16.645 --> 00:20:17.765

with the sales and with them,

451

00:20:18.545 --> 00:20:21.525

and that's moved us forward as a step change.

452

00:20:21.945 --> 00:20:23.765

So almost as something's being developed,

453

00:20:23.785 --> 00:20:27.085

as being tested at the same time using, um,

454

00:20:27.225 --> 00:20:28.645

agents within the test tool.

455

00:20:29.465 --> 00:20:34.445

So it's shrunk the lifecycle, right.

456

00:20:34.865 --> 00:20:37.165

Um, but without compromising on quality.

457

00:20:37.705 --> 00:20:38.965

And, and it's a bit odd.

458

00:20:38.965 --> 00:20:40.405

We've got AI testing ai,

459

00:20:40.965 --> 00:20:43.605

but it's hopefully it creates itself in a Yes.

460

00:20:43.705 --> 00:20:45.285

Honest way. Yeah, it does.

461

00:20:45.785 --> 00:20:47.245

Um, no, we've got some really strong

462

00:20:47.245 --> 00:20:48.765

guardrails around the testing ai.

463

00:20:48.765 --> 00:20:52.525

Perfect. Yeah. Um, but it lets us find things early. Mm.

464

00:20:52.585 --> 00:20:54.765

And it's, it's like the fail fast mentality,

465

00:20:54.865 --> 00:20:57.845

but we're failing fast, uh, than we would with humans.

466

00:20:58.425 --> 00:21:00.205

You know, we don't have to feed the ai,

467

00:21:00.205 --> 00:21:01.525

give it days off, let it sleep.

468

00:21:02.145 --> 00:21:03.405

So, you know, it's, it's good.

469

00:21:03.945 --> 00:21:06.805

That's fantastic. So taking kind of a step

470

00:21:07.315 --> 00:21:10.565

away from Bordeaux sync specifically, just in terms

471

00:21:10.625 --> 00:21:13.125

of the applications of these, these kinds

472

00:21:13.125 --> 00:21:17.525

of ideas on other elements within the insurance space, um,

473

00:21:17.595 --> 00:21:18.925

what else are you excited about?

474

00:21:18.925 --> 00:21:21.605

Like what's kind of next on the horizon for you? Um,

475

00:21:21.995 --> 00:21:24.925

Well, bores sync is one of our AI strands.

476

00:21:24.925 --> 00:21:28.525

We've done a lot of work around, um, ai, WhatsApp.

477

00:21:29.385 --> 00:21:31.805

So, and this is more in the general insurance space,

478

00:21:31.865 --> 00:21:34.645

but that's where, you know, if you have an insurance policy

479

00:21:34.645 --> 00:21:35.805

and let's say it's a motor policy

480

00:21:35.865 --> 00:21:39.645

and you wanna drive to Europe, you can use, um, WhatsApp

481

00:21:39.745 --> 00:21:43.725

to contact your insurance company, um,

482

00:21:43.825 --> 00:21:45.845

and it's got an AI agent that's powered there

483

00:21:45.845 --> 00:21:48.085

that can find your policy, it can find the exclusions.

484

00:21:48.105 --> 00:21:51.005

It can say if you are, um, covered to drive in Europe,

485

00:21:51.225 --> 00:21:53.485

if you're not, it can add it into your policy.

486

00:21:54.065 --> 00:21:56.765

It can raise the premium for it, for the, um,

487

00:21:58.235 --> 00:22:00.605

I've forgotten the technical term now, the endorsement.

488

00:22:00.605 --> 00:22:02.805

That's it. Um, and,

489

00:22:02.905 --> 00:22:05.325

and that, you know, there's a lot of, um,

490

00:22:07.035 --> 00:22:08.045

benefit in that.

491

00:22:08.365 --> 00:22:10.245

Probably not so much in the London market to be fair,

492

00:22:10.705 --> 00:22:13.285

but, you know, there's general insurances, life insurance,

493

00:22:13.705 --> 00:22:14.885

you know, so we're looking at that.

494

00:22:15.345 --> 00:22:17.125

Um, we've also done a lot with voice ai,

495

00:22:17.505 --> 00:22:18.645

um, for call centers.

496

00:22:19.505 --> 00:22:21.525

And the thing is, it's not to replace people.

497

00:22:22.275 --> 00:22:24.885

It's a lot of the very easy questions.

498

00:22:25.265 --> 00:22:27.925

Um, the voice AI can handle

499

00:22:28.145 --> 00:22:30.245

and it can have a conversa, it's got to the stage now,

500

00:22:30.245 --> 00:22:31.845

it can just have a natural conversation.

501

00:22:31.985 --> 00:22:34.965

And I guarantee if you've got someone to ring a call center

502

00:22:34.985 --> 00:22:36.365

and one number was a human

503

00:22:36.365 --> 00:22:37.405

and one number was ai,

504

00:22:37.405 --> 00:22:39.045

you couldn't tell the difference. Hmm.

505

00:22:39.515 --> 00:22:41.325

Yeah, we're doing a lot in kind

506

00:22:41.325 --> 00:22:43.205

of connected customer service as well.

507

00:22:43.675 --> 00:22:47.285

It's a really key area where I think we can improve the,

508

00:22:47.465 --> 00:22:50.365

the quality of interactions from the customer point of view.

509

00:22:50.385 --> 00:22:51.445

So that's really fantastic

510

00:22:51.475 --> 00:22:52.685

that you're kind of digging in there.

511

00:22:53.155 --> 00:22:56.565

What do you think the challenges that are, are kind

512

00:22:56.565 --> 00:22:58.605

of facing the insurance space?

513

00:22:58.635 --> 00:23:00.165

What, what do you think are the hurdles

514

00:23:00.165 --> 00:23:01.565

that might be slowing

515

00:23:01.565 --> 00:23:04.205

or stopping, um, these things from accelerating?

516

00:23:04.585 --> 00:23:07.765

Um, it, yes, uh, ironically the insurance

517

00:23:08.325 --> 00:23:09.405

industry is very risk averse.

518

00:23:10.185 --> 00:23:14.125

Um, I, I think it, it's one of those things, it's like,

519

00:23:14.145 --> 00:23:16.645

you know, when the internet first launched, you know,

520

00:23:16.745 --> 00:23:19.285

you had like senior, oh, we've gotta be on the internet,

521

00:23:19.285 --> 00:23:20.325

but what does it actually mean?

522

00:23:20.325 --> 00:23:22.485

And AI's, oh, we've gotta be using ai.

523

00:23:22.945 --> 00:23:26.925

And I think the problem is ai, as I said, is not magic.

524

00:23:27.465 --> 00:23:28.965

It needs structure, it needs to learn,

525

00:23:28.965 --> 00:23:30.165

it needs understanding, and it needs

526

00:23:30.165 --> 00:23:31.605

to be applied to a specific problem.

527

00:23:32.665 --> 00:23:35.165

Um, and it's finding out what that problem is.

528

00:23:35.505 --> 00:23:37.205

But obviously, you know,

529

00:23:37.205 --> 00:23:39.845

there's a lot in the fraud detection space it can do.

530

00:23:40.425 --> 00:23:42.485

Um, as I said, there's a lot in the sort

531

00:23:42.485 --> 00:23:44.245

of the general inquiry space it can do.

532

00:23:44.345 --> 00:23:46.565

We are doing a lot, lot of stuff with data with Res Inc.

533

00:23:47.105 --> 00:23:49.405

Um, it's finding the use case,

534

00:23:49.505 --> 00:23:51.485

and it, it's not necessarily the use case,

535

00:23:51.485 --> 00:23:52.725

doesn't have to be client facing.

536

00:23:52.905 --> 00:23:54.205

The use case could be internal,

537

00:23:54.785 --> 00:23:57.165

but it's using AI in a, in a

538

00:23:58.155 --> 00:23:59.885

thought about structured way.

539

00:24:00.345 --> 00:24:02.285

So it's actually solving a problem.

540

00:24:02.825 --> 00:24:04.805

And, but it's understanding, you know,

541

00:24:04.905 --> 00:24:06.205

how is it gonna solve that problem?

542

00:24:06.305 --> 00:24:07.605

What's your desired outcome?

543

00:24:07.985 --> 00:24:11.485

And I don't think it's necessarily restricted to insurance.

544

00:24:11.795 --> 00:24:14.685

It's, you know, it's emerging, it's new technology.

545

00:24:14.865 --> 00:24:18.005

And I sort of, I sort of asked chat sheet bt,

546

00:24:18.385 --> 00:24:20.765

but um, you know, like in 2024

547

00:24:22.325 --> 00:24:25.845

globally there was 16 billion pounds a month spent on AI

548

00:24:25.845 --> 00:24:28.685

development, which is a phenomenal, and b unsustainable.

549

00:24:29.665 --> 00:24:31.245

So at some point it's gonna slow down,

550

00:24:31.305 --> 00:24:32.805

and I think it's gonna get to the point

551

00:24:32.805 --> 00:24:36.845

where people start thinking, okay, the shiny new toy, great,

552

00:24:36.945 --> 00:24:39.005

now let's, how do we practically apply it?

553

00:24:39.385 --> 00:24:41.525

So I don't think it's, there's a reluctance

554

00:24:41.525 --> 00:24:43.085

to embrace it within the insurance industry.

555

00:24:43.315 --> 00:24:46.365

It's finding the specific way do it.

556

00:24:46.365 --> 00:24:47.925

And obviously the other way is ai,

557

00:24:48.235 --> 00:24:49.365

it's not cheap at the moment.

558

00:24:49.365 --> 00:24:52.645

Mm-hmm. So, you know, there, there's a considerable amount

559

00:24:52.645 --> 00:24:54.565

of investment, whether it's a prototype

560

00:24:54.625 --> 00:24:58.165

or it's buying a product and you wanna see a return on

561

00:24:58.165 --> 00:24:59.885

that investment as any company wants to

562

00:25:00.485 --> 00:25:02.485

A thousand percent, that's, we'd spend a lot

563

00:25:02.485 --> 00:25:05.165

of time working with our clients to find the right use cases

564

00:25:05.395 --> 00:25:07.285

that will generate that ROI.

565

00:25:07.355 --> 00:25:09.925

Yeah. And any kind of tips that you would have as someone

566

00:25:09.925 --> 00:25:11.885

that's successfully managed

567

00:25:11.905 --> 00:25:14.405

to get the investment on something that, um,

568

00:25:14.545 --> 00:25:15.805

you thought would be valuable?

569

00:25:16.405 --> 00:25:17.765

I, I think it's spend time on,

570

00:25:18.265 --> 00:25:19.805

so there's two ways to do it.

571

00:25:19.865 --> 00:25:21.925

You can either set up an r and d team

572

00:25:22.025 --> 00:25:24.645

and accept the fact they're gonna burn some cash and,

573

00:25:25.265 --> 00:25:28.005

but go go to the business and ask them what's the problem?

574

00:25:28.945 --> 00:25:30.565

You know, if we could solve one thing

575

00:25:30.565 --> 00:25:32.805

that would make your life better, what is it now?

576

00:25:32.805 --> 00:25:34.605

Maybe AI can help, maybe it can't,

577

00:25:34.665 --> 00:25:38.805

but it's spending that time to validate the, the use case

578

00:25:38.985 --> 00:25:40.805

or the problem, and also knowing

579

00:25:40.835 --> 00:25:42.325

what you want the outcome to be.

580

00:25:42.785 --> 00:25:46.165

So you can't just give AI a problem and ta, right? Yeah.

581

00:25:46.165 --> 00:25:48.285

There's the answer. You know, it's gotta, you know,

582

00:25:48.285 --> 00:25:51.645

it learns, it has to know what you want it to achieve.

583

00:25:52.145 --> 00:25:54.605

And, you know, sometimes there's nothing wrong

584

00:25:54.725 --> 00:25:56.125

with trial trying and failing,

585

00:25:56.665 --> 00:25:59.365

but there's also that bit of human ingenuity

586

00:25:59.625 --> 00:26:01.525

and review to say, actually,

587

00:26:01.665 --> 00:26:04.565

is this the right technology to solve that problem?

588

00:26:05.675 --> 00:26:06.885

Yeah, absolutely.

589

00:26:07.225 --> 00:26:08.565

And why would you say it's important

590

00:26:08.565 --> 00:26:11.445

to have a partner like transparency, would you say?

591

00:26:12.635 --> 00:26:14.805

Well, okay, well, okay, I'll be careful here.

592

00:26:15.225 --> 00:26:17.685

Um, you've gotta understand your own

593

00:26:17.685 --> 00:26:18.725

limitations and your own.

594

00:26:18.905 --> 00:26:20.205

So, you know, if, if you take it,

595

00:26:20.545 --> 00:26:22.645

that's a very broad brush term

596

00:26:22.665 --> 00:26:24.765

and covers a lot of different things.

597

00:26:25.625 --> 00:26:27.725

So if you want to accelerate

598

00:26:28.345 --> 00:26:30.925

and you want to sort of be at the forefront

599

00:26:30.925 --> 00:26:33.765

or ahead of the curve, if you don't have those skills in

600

00:26:33.765 --> 00:26:36.125

house, it's, even if you went

601

00:26:36.125 --> 00:26:37.645

and hired people, it's gonna take time.

602

00:26:37.715 --> 00:26:40.285

It's gonna be delay. If you're trying to upskill your,

603

00:26:40.555 --> 00:26:42.045

your incumbent teams, you know,

604

00:26:42.045 --> 00:26:44.205

you can't forget about your current customers,

605

00:26:44.205 --> 00:26:45.805

your applications, your BAU.

606

00:26:46.705 --> 00:26:49.805

So, you know, we looked at it as a way of an accelerant,

607

00:26:50.135 --> 00:26:54.005

accelerant accelerator, and, um, that, you know,

608

00:26:55.225 --> 00:26:57.605

and you, because, you know, you have

609

00:26:58.305 --> 00:27:00.045

key competencies in this area

610

00:27:00.265 --> 00:27:03.765

and also, you know, you do work with Microsoft

611

00:27:03.945 --> 00:27:05.725

and your Microsoft technology partner,

612

00:27:05.735 --> 00:27:07.005

we're on a Microsoft stack.

613

00:27:07.385 --> 00:27:08.845

So there's a lot of synergies there

614

00:27:08.845 --> 00:27:10.765

and a lot of understanding just out of the box.

615

00:27:11.065 --> 00:27:13.085

You don't have to become familiar with our platforms,

616

00:27:13.085 --> 00:27:15.365

you don't have to become familiar with Microsoft technology,

617

00:27:15.705 --> 00:27:18.525

you know, what's coming out from Microsoft, you know,

618

00:27:18.675 --> 00:27:22.325

what can potentially enhance or be adapted and,

619

00:27:23.025 --> 00:27:24.645

and sort of keep the momentum,

620

00:27:24.785 --> 00:27:26.765

but sort of also then get ahead of the curve.

621

00:27:27.345 --> 00:27:28.645

And I think that's really important

622

00:27:28.755 --> 00:27:30.325

because, you know, yes,

623

00:27:30.325 --> 00:27:31.885

we could have built this from scratch ourselves,

624

00:27:31.905 --> 00:27:34.765

but it would've taken a hell of a lot of longer than it has.

625

00:27:35.065 --> 00:27:37.445

And we would be saying, showing someone a prototype

626

00:27:37.445 --> 00:27:38.845

and saying, oh yeah, in two years

627

00:27:38.975 --> 00:27:40.205

we'll come back and you can have it.

628

00:27:40.515 --> 00:27:42.965

Exactly. Yeah, I think there's a big element, uh,

629

00:27:42.965 --> 00:27:45.805

especially within it of, you know, keeping the lights on,

630

00:27:45.935 --> 00:27:47.245

which is really critical.

631

00:27:47.245 --> 00:27:48.405

Yeah. You know, you're, you're dealing with

632

00:27:48.605 --> 00:27:49.885

business critical functions Yeah.

633

00:27:50.035 --> 00:27:51.365

Just on a day-to-day level.

634

00:27:51.625 --> 00:27:54.045

So yeah, finding someone that can help you accelerate

635

00:27:54.185 --> 00:27:56.005

as you keep the lights on Yeah.

636

00:27:56.025 --> 00:27:58.365

At the same time can be really useful, I think. Yeah.

637

00:27:58.585 --> 00:28:00.925

And, and I think the other big benefit is we have involved

638

00:28:00.925 --> 00:28:03.285

our IT team, so it's not like doing

639

00:28:03.285 --> 00:28:04.405

it in a darkened basement.

640

00:28:04.705 --> 00:28:07.205

So even though they're not, I guess, um,

641

00:28:07.925 --> 00:28:10.805

participating actively, they're in all the meetings are,

642

00:28:10.805 --> 00:28:13.005

they're in the conversations, you know, there's

643

00:28:13.605 --> 00:28:16.245

learning way osmosis and that sort of thing.

644

00:28:16.305 --> 00:28:19.205

So it's, you know, that's another beneficial and,

645

00:28:19.205 --> 00:28:20.805

and transfer quite happy to work that way.

646

00:28:21.025 --> 00:28:22.965

And it, and it's also like in a, an odd way,

647

00:28:22.965 --> 00:28:24.045

you don't wanna be doing this forever.

648

00:28:24.265 --> 00:28:26.925

You wanna some point say, no, it's yours now. Go, you

649

00:28:26.925 --> 00:28:27.925

Know? Yeah, exactly. Yeah.

650

00:28:27.925 --> 00:28:29.245

That knowledge transfer piece

651

00:28:29.245 --> 00:28:30.845

of it is definitely built into the project.

652

00:28:31.185 --> 00:28:34.645

Um, and then, yeah, thinking about the future of BRE Inc.

653

00:28:35.065 --> 00:28:37.685

Um, what's kind of next on the, on the list?

654

00:28:37.985 --> 00:28:39.005

What's, what's to come?

655

00:28:39.435 --> 00:28:41.325

Well, we have a very long backlog,

656

00:28:41.925 --> 00:28:42.925

A long List. It's long

657

00:28:42.925 --> 00:28:46.005

list. So bres syncs standalone application,

658

00:28:46.145 --> 00:28:48.285

and it's des and it's purposely designed that way,

659

00:28:48.425 --> 00:28:49.765

and it's designed to work

660

00:28:49.835 --> 00:28:51.565

with any boardroom management solution.

661

00:28:51.995 --> 00:28:55.205

However, we have our own, so we want to give added value

662

00:28:55.225 --> 00:28:56.845

to the customers who are using ours

663

00:28:57.225 --> 00:29:00.845

and also start selling it as potentially a, a package.

664

00:29:01.585 --> 00:29:04.885

So it does what it says on the tin, it's good.

665

00:29:04.885 --> 00:29:06.645

So we've, we've productionized it,

666

00:29:06.645 --> 00:29:10.445

or it's almost productionized, um, from the trial version.

667

00:29:11.225 --> 00:29:13.565

And now we're looking at how can we enhance

668

00:29:13.595 --> 00:29:15.405

what AI does, how can it learn further?

669

00:29:15.585 --> 00:29:19.205

So we're integrating with a small lie to our

670

00:29:19.805 --> 00:29:23.845

wardrobe management solution tied, um, or coupling it,

671

00:29:24.065 --> 00:29:26.365

and that leverages greater functionality.

672

00:29:26.585 --> 00:29:29.645

So, you know, as I said earlier, if data's missing

673

00:29:29.745 --> 00:29:34.085

or it can't find, you know, the correct correction

674

00:29:34.105 --> 00:29:37.485

for bad data, it has to go back and ask someone.

675

00:29:38.385 --> 00:29:40.005

But without, you know, coupling it

676

00:29:40.005 --> 00:29:41.605

with our board management solution, a lot

677

00:29:41.605 --> 00:29:45.125

of those answers already reside in our solution so it can go

678

00:29:45.125 --> 00:29:47.685

and look at our solution and find the answers.

679

00:29:48.105 --> 00:29:50.125

So then that speeds up the process.

680

00:29:51.065 --> 00:29:53.725

And, you know, without going into insurance just

681

00:29:53.725 --> 00:29:54.925

for a second, you know,

682

00:29:55.415 --> 00:29:57.125

think something called return premium.

683

00:29:57.385 --> 00:29:59.325

So you've got a policy, you've paid your premium,

684

00:29:59.465 --> 00:30:01.645

you cancel partway through, you owed some money back.

685

00:30:02.165 --> 00:30:06.365

Normally, um, those refunds come through on the bojo,

686

00:30:06.585 --> 00:30:08.965

but you've got no idea if it's the right amount.

687

00:30:09.585 --> 00:30:12.605

Um, so with bojo sink, when it's coupled, it can go

688

00:30:12.605 --> 00:30:15.005

and see how much the original premium is, what's been paid,

689

00:30:15.625 --> 00:30:18.885

is the refund the right amount?

690

00:30:19.225 --> 00:30:21.885

And if it's not, then it can identify someone, uh,

691

00:30:22.325 --> 00:30:25.165

identify someone, identify and tell someone that it's wrong.

692

00:30:25.345 --> 00:30:28.325

So there's, there may not be step changes and,

693

00:30:28.425 --> 00:30:31.685

and functionality per se, but that step changes

694

00:30:31.865 --> 00:30:34.165

and what it gives to the end users

695

00:30:34.265 --> 00:30:36.965

and how it speeds up their process

696

00:30:37.185 --> 00:30:40.445

and how it alerts them to issues, um,

697

00:30:40.925 --> 00:30:42.445

a lot earlier in the lifecycle.

698

00:30:42.985 --> 00:30:44.885

Now with delegated, it's,

699

00:30:44.885 --> 00:30:46.245

the problem is you're always getting information

700

00:30:46.245 --> 00:30:48.405

after the event, so it may have already happened,

701

00:30:48.425 --> 00:30:50.925

but they're finding out there's been an error sooner.

702

00:30:51.465 --> 00:30:52.605

And that's really valuable

703

00:30:52.605 --> 00:30:55.325

to firms when you're talking about real money.

704

00:30:55.755 --> 00:30:58.085

Yeah, exactly. That sounds really, really powerful.

705

00:30:58.385 --> 00:31:00.525

And, um, definitely things that, my,

706

00:31:00.625 --> 00:31:04.165

my dad's work in the reinsurance industry is kind of echoed.

707

00:31:04.235 --> 00:31:05.765

He's always talking about premiums

708

00:31:05.865 --> 00:31:08.045

and how to make sure that everything's Yeah.

709

00:31:08.765 --> 00:31:10.165

Accurate. Ultimately, it's really,

710

00:31:10.165 --> 00:31:11.285

really important decisions.

711

00:31:11.515 --> 00:31:13.125

Yeah. And, and we are looking at, you know, we

712

00:31:13.185 --> 00:31:15.325

as going back to, I said we are looking at various use

713

00:31:15.325 --> 00:31:16.805

cases, some of them might fly, some not,

714

00:31:16.805 --> 00:31:18.685

but also around what can we do around fraud.

715

00:31:19.225 --> 00:31:22.485

Now, again, delegated the fraud's already happened,

716

00:31:23.425 --> 00:31:27.045

but if AI can be used to help detect it or anomalies,

717

00:31:27.145 --> 00:31:30.725

or especially around claims borow, um, especially

718

00:31:30.725 --> 00:31:33.565

around catastrophic events, which I appreciate aren't great,

719

00:31:33.785 --> 00:31:34.845

but, you know, but

720

00:31:35.265 --> 00:31:36.085

You prepare. Yeah,

721

00:31:36.085 --> 00:31:37.085

Exactly. Yeah. And,

722

00:31:37.085 --> 00:31:38.565

and, you know,

723

00:31:39.105 --> 00:31:40.805

and then what are those learnings that we can take

724

00:31:40.805 --> 00:31:41.805

to our other towers?

725

00:31:42.065 --> 00:31:43.765

Um, so we have fraud solutions.

726

00:31:43.775 --> 00:31:45.525

We're looking at what can AI do around there

727

00:31:45.525 --> 00:31:48.925

that's more real time, it's more at, um, first notification

728

00:31:48.925 --> 00:31:51.045

of loss, those sorts of things.

729

00:31:51.225 --> 00:31:54.725

So it's, it, it's great in the way we can,

730

00:31:54.745 --> 00:31:56.165

we can build something functionally

731

00:31:56.165 --> 00:31:58.645

that we know addresses a real problem in the market,

732

00:31:58.745 --> 00:31:59.765

we can enhance it,

733

00:32:00.145 --> 00:32:02.165

but then how do we move that from sort of

734

00:32:02.945 --> 00:32:05.045

moving it further to the left mm-hmm.

735

00:32:05.105 --> 00:32:07.125

To the point where it's happening real time

736

00:32:07.305 --> 00:32:10.845

and it's giving those insights and warnings

737

00:32:10.845 --> 00:32:13.325

and capabilities to people that, you know,

738

00:32:13.535 --> 00:32:14.885

maybe they wouldn't pick up.

739

00:32:15.395 --> 00:32:16.805

Yeah, exactly. That was kind

740

00:32:16.805 --> 00:32:19.565

of my next question was if we gave you a crystal ball,

741

00:32:19.905 --> 00:32:21.845

you know, what would the insurance industry

742

00:32:21.845 --> 00:32:23.045

look like in a couple of years?

743

00:32:23.105 --> 00:32:24.405

But I think you're talking about that more

744

00:32:24.475 --> 00:32:25.525

real time activation.

745

00:32:26.235 --> 00:32:28.605

Yeah. And, and I think this is, this is gonna send,

746

00:32:29.085 --> 00:32:30.525

I think the real big change,

747

00:32:30.905 --> 00:32:34.925

and now I'll stand to be corrected here, is tech,

748

00:32:34.925 --> 00:32:36.125

the way technology is moving,

749

00:32:36.125 --> 00:32:38.685

and not technology per se for the insurance industry,

750

00:32:39.385 --> 00:32:42.485

but if I take, you know, automation, so driverless trains,

751

00:32:42.485 --> 00:32:46.245

driverless cars, driverless uh, planes,

752

00:32:46.555 --> 00:32:49.925

well pilotless planes, that is, you know, and,

753

00:32:49.925 --> 00:32:51.685

and this is probably, this is staying away from London

754

00:32:51.685 --> 00:32:55.005

market, but you know, today people take insurance out

755

00:32:55.345 --> 00:32:59.085

to give them some security in case an event happens.

756

00:32:59.345 --> 00:33:01.245

And if it happens, normally insurance is

757

00:33:01.245 --> 00:33:04.085

to put them back in the place they were before the event.

758

00:33:04.865 --> 00:33:06.325

But if they take driverless cars

759

00:33:07.025 --> 00:33:09.605

and it's in an accident, who's liable?

760

00:33:10.145 --> 00:33:12.405

The person who owns the car, the person who made the car,

761

00:33:12.405 --> 00:33:13.645

the person who supplied the software

762

00:33:14.115 --> 00:33:15.365

Yeah. Programmed the car,

763

00:33:15.545 --> 00:33:16.605

The car program, the car.

764

00:33:16.945 --> 00:33:18.885

And so, you know, people will still have motor insurance,

765

00:33:19.025 --> 00:33:23.565

but the content of that may change as technology adapts

766

00:33:23.585 --> 00:33:27.605

and the way these, you know, assets, things mm-hmm.

767

00:33:27.755 --> 00:33:30.405

Operate. And then the other one we were talking about is

768

00:33:30.405 --> 00:33:32.605

there's this big fear that AI is gonna take jobs.

769

00:33:33.915 --> 00:33:36.455

So what are the new products?

770

00:33:36.475 --> 00:33:39.135

And if I give anyone an idea, I want commission here. Yeah.

771

00:33:39.275 --> 00:33:41.655

Um, um, what new product, you know,

772

00:33:41.655 --> 00:33:43.695

are you gonna have such things like as AI insurance?

773

00:33:43.845 --> 00:33:45.895

Yeah, you are, but is a human gonna be able

774

00:33:45.895 --> 00:33:48.575

to take AI insurance out in case they lose their job to ai?

775

00:33:50.485 --> 00:33:52.905

So, you know, it's fascinating. Yeah. Yeah.

776

00:33:52.905 --> 00:33:56.625

It's not necessarily what's technology gonna do directly

777

00:33:56.625 --> 00:33:59.625

for insurances, what's the indirect impact it's gonna have?

778

00:34:00.325 --> 00:34:04.105

And also, you know, using the, the real time information,

779

00:34:04.885 --> 00:34:09.265

um, predictive models, taking data from disparate sources,

780

00:34:10.015 --> 00:34:12.785

putting it all together, giving a view, and,

781

00:34:12.885 --> 00:34:13.905

and, you know,

782

00:34:13.905 --> 00:34:15.825

people have always gone about predictive analytics,

783

00:34:15.835 --> 00:34:17.825

which has always been kind of like nice to have,

784

00:34:17.845 --> 00:34:18.985

but I think getting to the stage

785

00:34:18.985 --> 00:34:21.785

where the prediction can probably come a lot more accurate

786

00:34:21.895 --> 00:34:24.905

than some of the older technology purported to be.

787

00:34:25.705 --> 00:34:27.985

Absolutely. Yeah. I just was in, um, San Francisco

788

00:34:27.985 --> 00:34:30.025

for the Microsoft Ignite Conference,

789

00:34:30.165 --> 00:34:32.665

and yeah, there's driverless cars zooming around.

790

00:34:33.005 --> 00:34:34.985

Um, I was trying to make eye contact

791

00:34:35.055 --> 00:34:36.505

with one when I was crossing the street

792

00:34:36.525 --> 00:34:37.705

to make sure everything was good

793

00:34:37.885 --> 00:34:39.865

and not sure that I could get that back.

794

00:34:40.005 --> 00:34:41.425

So yeah, there's a lot of kind of

795

00:34:41.615 --> 00:34:44.185

what is the downstream application from an insurance

796

00:34:44.185 --> 00:34:46.025

standpoint, what does that mean for, for all

797

00:34:46.025 --> 00:34:47.225

of us, really Fascinating.

798

00:34:47.805 --> 00:34:49.305

Um, and then, yeah, the data all

799

00:34:49.305 --> 00:34:50.785

coming together in near real time.

800

00:34:51.025 --> 00:34:53.385

I mean, you can even bring in weather patterns Yeah.

801

00:34:53.405 --> 00:34:55.985

To determine what types of policies people need.

802

00:34:56.365 --> 00:34:58.945

Um, so yeah, I think there's, there's a lot ahead and,

803

00:34:58.945 --> 00:35:00.105

and it's just really exciting

804

00:35:00.125 --> 00:35:02.385

how you've already made a start with something

805

00:35:02.385 --> 00:35:03.945

that has real impact for your clients.

806

00:35:04.165 --> 00:35:06.185

So thanks for sharing. That's all right.

807

00:35:06.405 --> 00:35:08.305

Any last, um, comments or,

808

00:35:08.365 --> 00:35:11.185

or things that you'd wanna share with our podcast listeners?

809

00:35:11.295 --> 00:35:13.185

It's exciting times around technology.

810

00:35:13.625 --> 00:35:18.265

I think, um, you know, with, with all things you, you have

811

00:35:18.265 --> 00:35:20.305

to approach it with an ear of caution.

812

00:35:21.165 --> 00:35:23.425

Uh, don't blindly trust ai, um,

813

00:35:23.965 --> 00:35:25.865

and make sure you're using it in the right way

814

00:35:25.965 --> 00:35:27.825

for the right challenge.

815

00:35:28.085 --> 00:35:29.305

Mm. You know, um,

816

00:35:29.965 --> 00:35:33.425

and I couldn't recommend more highly than, you know,

817

00:35:34.285 --> 00:35:37.205

if you don't have those skills, find, find companies that do

818

00:35:37.265 --> 00:35:40.565

and partner with them, but collaboratively, um,

819

00:35:40.835 --> 00:35:44.485

because you will get a lot more outta the relationship, um,

820

00:35:44.755 --> 00:35:49.125

than a traditional customer vendor type arrangement.

821

00:35:49.705 --> 00:35:54.325

Um, and you know, you, you will be able to hopefully,

822

00:35:54.905 --> 00:35:57.285

you know, move things forward at a, at a great race

823

00:35:57.305 --> 00:35:59.725

and loss with knots, with a par a positive outcome.

824

00:36:00.725 --> 00:36:02.325

Absolutely. Well, thanks so much Reese.

825

00:36:02.325 --> 00:36:04.965

Really appreciate all of your thoughts and leadership here.

826

00:36:04.965 --> 00:36:07.285

It's been fantastic partnering with you on this.

827

00:36:07.385 --> 00:36:08.405

And really, yeah.

828

00:36:08.405 --> 00:36:10.125

The collaborative approach has been fantastic.

829

00:36:10.865 --> 00:36:14.165

Um, thanks to all of our leaders who are, thanks to all

830

00:36:14.165 --> 00:36:17.045

of our listeners, um, who have listened to the podcast,

831

00:36:17.235 --> 00:36:19.885

hope you've gotten some insights from Reese,

832

00:36:19.945 --> 00:36:24.005

how he's really brought AI into, um, reality

833

00:36:24.185 --> 00:36:26.605

for his customers with a cross-functional team.

834

00:36:27.105 --> 00:36:28.285

Um, hopefully some tips

835

00:36:28.285 --> 00:36:31.125

and tricks that you can use as you go on your AI journey.