The innovation myth: why big insurers can move faster than you think, with Dominic Wate

Large organisations are often seen as too slow, too cautious and too weighed down by process to innovate. But is that actually true? In this episode, Georgie Simister is joined by Dominic Wate, Head of IT Enablement and Acceleration at TMHCC International, to challenge that assumption. They explore how governance and compliance can be a foundation for experimentation rather than a barrier to it, what safe AI adoption looks like in practice at a global insurer, and why quantum computing may be the next thing on everyone's radar whether they like it or not.

Creators and Guests

Host

Georgie Simister

Georgie is the Commercial Partnerships Lead at Artificial, a London-based insurance technology partner to global insurers and brokers.

What is Insurance Technology: Fact or Fiction??

Host Georgie Simister speaks with key insurance industry figures to separate fact from fiction in a world of insurance, debunk industry myths and explore the game-changing innovations shaping its future.

Get in touch to suggest our next guest: https://artificial.io/contact/

Georgie: 00:01

Big organisations are often seen as slow, complex and held back by bureaucracy. But does that really need to stop innovation? Today we are exploring how large insurers can move quickly, embrace AI and still operate safely and at scale. Today I am joined by Dominic Wate, who is Head of IT Enablement and Acceleration at TMHCC International, where he focuses on helping a global insurer get more value from technology by connecting innovation with real business needs. His work centres on enabling teams to experiment with emerging technologies like AI in practical and responsible ways and is promoting IT as a proactive, insight driven partner to the business.

Georgie: 00:46

He has strong interest in areas such as quantum computing and the future of enterprise technology. Dominic brings a thoughtful and grounded perspective on how large organisations can evolve and adapt in fast moving landscapes. Dominic, welcome to Insurance Technology.

Dominic: 01:02

Thank you. Thanks for having me on today.

Georgie: 01:04

No worries at all. Your introduction was fantastic. I feel like we need more people in insurance that have similar kind of passion and responsibilities of you. So, yeah, really looking forward to this discussion. Let's start with the big one.

Georgie: 01:19

Does bureaucracy actually get in the way of innovation innovation in large organisations? Or is it too simplistic?

Dominic: 01:25

Too simplistic is the way to put it. I love and hate the word bureaucracy. I love it because I love words. If we think about what bureaucracy actually means, you can compare it to autocracy and democracy. And my favorite one lately is kakistocracy, which is a country run by the worst people, as an indication there.

Dominic: 01:42

But these words, they are autocracy is run by a dictator, democracy is run by the people. The word bureaucracy, it implies that it's run by the paperwork. So I think the philosophy has to change there. It's not that you have these bureaucracy blockers. What you have is a series of governance tools and regulations and policies which create a safe environment for innovation, for anything really, for any especially in the IT space.

Dominic: 02:13

So for me, it's about flipping that word on its head and shuffling it away when it comes to enabling innovation. Innovation is not bombarded or blocked by these bureaucratic needs, it's supported by your compliance and your governance.

Georgie: 02:26

So, okay, so fundamentally removing that word to enable innovation then.

Dominic: 02:32

Yeah. I mean, well, the connotation, what people associate bureaucracy with that red tape and that paperwork. What's great about being in a large organisation is it's got the money to invest in these guardrails, these safety nets that enable you to innovate in a safe space.

Georgie: 02:48

Yeah. Well, so your role, you'll get quite a lot of insight from the underwriters. And from conversations that you're having today, what challenges are the underwriters most commonly coming to you to solve?

Dominic: 02:59

So I'd say it's not even just our underwriters, it's the whole industry. If you look at simple policies that you'd get for your car or your home insurance, you go on a comparison website, you fill in your registration and your mileage and you've got your car insurance quotes. Specialty insurance is not like that. You could have thousand page documents. And this is where you don't take away the expertise that comes from being an underwriter, their experience, their knowledge, their talent in understanding the risk and pricing accordingly and thinking about their long term strategy.

Dominic: 03:36

What you do is you accelerate them by using the AI tools, other innovative tools to speed up the process that frankly just take a human a long time to do. That's where I'm seeing across the whole industry, across the insurance industry, the acceleration of the sort of repetitive deep reading tasks that computers can take away from them.

Georgie: 03:57

On that topic of more people coming to you for kind of IT guidance, when people think about digital transformation, what area is it that they are looking at it? So from my lens, I think of digital transformation, and in my role, I see it purely as being underwriter focus and what you're saying. But is that common across the board? Or when people think of digital transformation, are they more thinking of back office processes at the moment?

Dominic: 04:23

You're absolutely right. I'd say underwriting definitely has its use cases and there are conversations going on, as I say, across the industry. But when you look at your claim space and your operation space, that's really where efficiencies can be had as well. You've got to be careful. My bosses, they talk about the ambulance problem, which is essentially you can speed up the process of dispatching and returning ambulances to the hospital.

Dominic: 04:47

But if you haven't sped up your emergency room processes, they're just going to bottleneck there. So you've to think about it end to end. And therefore, these questions around underwriting, claims, operations, you need to look at it holistically. But that said, operations dealing with the administrative work, the paperwork after the fact, there's enormous efficiencies to be had with these large documents, these big processes that can really be accelerated. Again, not taking away from the human aspect skills because AI is very fast.

Dominic: 05:21

But honestly, it's still pretty dumb. It can do things in more and more interesting ways. But the human is where the real power is. We just need to speed them up.

Georgie: 05:31

There's a growing narrative that if you're not using AI right now, you're being left behind. How do you support teams who want to use AI quickly? And when there's a lot of talk around safe AI adoption, what does safe actually mean in practice to an organisation like yours?

Dominic: 05:47

So there's a couple of things. Firstly, let's jump back to that dreaded bureaucracy. It's not bureaucracy. You can create a safe environment for innovating by giving people the guardrails. You build a sandbox, essentially, which says as long as you follow these rules, you are free to experiment.

Dominic: 06:04

And that experiment is really where the real value. We're talking about, it's essentially science here. I have a business value hypothesis. I want to prove that. Where can I do it?

Dominic: 06:15

So any large organisation can utilise their regulations, governance, their compliance policies, establish a safe sandbox for them to play, let them play. Once you've reached that point where you've proven your business value hypothesis, then you can move into your standard approach of sitting back and designing and working out your value stream, engineering and delivery in a more controlled fashion. Because you've very quickly been able to prove your point. So the sandbox is what's key here. You can speak to your teams and say, I want you to do more with AI.

Dominic: 06:48

And they can ask a chatbot question. That's fine. It's still useful, it's interesting and it'll probably speed them up. But that real sort of transformative value is gonna come from these bigger experiments. And you need that safe space to do that in.

Georgie: 07:02

How long does it take you or take the team to go from, they have an idea to sandbox then through to production? What kind of timeframes are you typically talking?

Dominic: 07:10

Very variable. It depends. So if we talk about I know we're talking about innovation today, but let's not kill ourselves, AI is a huge topic. There's different types of AI. There's me asking ChatGPT a question.

Dominic: 07:22

Or there's me customising a conversation, a custom GPT with ChatGPT. And these things are the personal AI. I have a very immediate productivity need. I can have a conversation and get what I need. But on the other end of the spectrum, you have this idea of this fully agentic orchestrated solution where I can give it a bunch of complicated information, it has access to a lot of context, a lot of powerful backend solutions and it can put that together for you and act on your behalf with the human in the loop of course.

Dominic: 07:57

So to answer your question, it's hugely variable as to the time you're taking from inception of an idea to experimenting to delivery. But to give an example, am happy to share, we recently looked into creating a chatbot assistant. So not just a one on one conversation ad hoc but a knowledge augmented solution for a specific use case. And the team delivered that in three days. Took all the information they had and built the Orchestrate chatbot and delivered it.

Dominic: 08:31

Now, what's interesting there to look back on the history is the first time that was done, it was a few months and then it was a few weeks and then it was a few days. And that acceleration comes from giving people the ability and the safety to experiment and learn and grow and develop these patterns. So that approach is it sells itself. Essentially, that return on investment of you giving people the space to innovate shows up in the final product.

Georgie: 08:59

A lot of what we're seeing, you can no longer go in and just do a pitch and tell somebody that you have this great thing. The underwriters need to get their paws on it. And I completely get that because it's their day to day job and they need to see that it works for them because there are so many nuances. When you're doing a pilot and you're proving the concept of the theory and what they're trying to achieve with technology, once you've done it with one line of business, does that then typically mean that that is a concept or that's been proven across the board? Or do you have to run multiple different pilots for all the different lines of business for it to become more trusted?

Georgie: 09:37

How does that look for you?

Dominic: 09:38

It's a bit of combination. I think my perception is it leans towards the former. You know, there's a lot of reusability baked into this and that's kind of like the example I gave you with the accelerated time to deliver because we've learned from the previous incarnations, you can reuse patterns, you can reuse solutions, and you've delivered a working solution for that new line of business. In reality, there's always going to be cases where you have these nuances. Insurance has some very common principles.

Dominic: 10:07

We deal with policies. We deal with claims. But a special line of business, they have their own particular needs. So it comes down to how much is reusable, how much is unique to those lines of business and also how much trust people have. It's early days for the whole industry, for the whole market, beyond insurance as well.

Dominic: 10:28

It's early days for the trust and faith that we have in AI and it makes sense to spend some time proving that. And again, it comes down to having a safe environment to do that in.

Georgie: 10:41

So to summarise then, a pilot that has worked well, it's getting it, it's getting the underwriters have come to you with problem that they want to solve. It's getting them into a sandbox environment to prove the concept. Is there anything else that you think in the pilot phase that has been really successful?

Dominic: 10:58

Yeah, so I talked earlier about this business value hypothesis, I think because there's a risk of new toys, new tools, where you just sort of go off and you start playing around and you forget what you're doing it for. We are a business. So having a clear business value hypothesis right at the offset enables you to focus on what's important to the stakeholder. And I mentioned it being a scientific experiment because you set out, I believe that this value will be delivered. Let me test that.

Dominic: 11:32

Now I do like to explain to people that that's probably 90% to 95% of the pilots or POCs that we explore. You do want to leave some room in a healthy organisation, Maybe it's as little as 5% for what we call the research question, where there is a new toy. I just want to see what this can do. That tends to be your techie teams for the most part, but not exclusively. So you want to have a bit of room there.

Dominic: 11:56

But the real success of being able to turn an experiment into a valid and valuable deliverable comes from staying from the beginning. This is what I think it's gonna do for me.

Georgie: 12:08

It leads me on nicely. The phrase FOMO is something that we are not shy of using in insurance and there's very much different hype cycles that we go through every time you go to an event, it's a new hype cycle that we're talking about. And you've kind of touched on it then, but how do you actually distinguish between a trend that you think that will fade and those ones that are worth serious investment?

Dominic: 12:31

Honestly, that's really tough. We have to go based on our history. We have to look at all those trends in the past and you have to use your gut sometimes. People around us have been involved in those various hype cycles so you start to see patterns and signs. The FOMO thing is interesting because the way I look at it, it's not so much a fear of missing out, it's kind of fear of being left behind.

Dominic: 12:53

So you don't want to just dismiss things. Oh, it's just a hype, I'll just move on. You've got to take some time and dedicate some research to that so that you can be in the best position to capitalise on it if necessary. But yeah, in terms of is this going to go away? Very, very difficult to predict.

Dominic: 13:16

If we could predict the future, we'd all be billionaires. And I'll keep coming back to this. But if you're giving people a safe and rapid place to experiment, you can very quickly determine if it's value for you. If you are saying to people, well, you've got to be absolutely sure because what you're asking for is a nine month, million pound project, then you'll either waste money on a hype, which is gonna go away, or you won't do it and you'll miss out.

Georgie: 13:42

You'll kill that kind of inquisitive nature of people.

Dominic: 13:45

Yeah, exactly.

Georgie: 13:47

With there being so many emerging trends that come through, what helps an organisation stay disciplined to avoid spreading themselves so thin across so many different emerging technologies?

Dominic: 13:58

I think the standard principles or practises of any organisation that of have they could be applied to innovation, they're applied to everything. Talk about spreading yourself too thin. Any given year, you might say, Well, I have these 300 projects I'd love to do, I can't do them all. It's just recognising that you have a limited amount of resources, money being the obvious factor, you have to prioritise. And prioritising a project or prioritising an investigation, some research and innovation, it's the same principle.

Dominic: 14:31

Now you might have enough information, enough research out there and some of that FOMO perhaps to say, well, I'm going to invest a bit more in that because I don't want to risk missing out on the opportunities. But it's still about prioritisation.

Georgie: 14:44

Do TMHCC International do anything in terms of like, do you have a full innovation arm that is purely looking at innovative products? And then do you have a separate one that's looking for innovative new products internally?

Dominic: 14:59

In terms of technology?

Georgie: 15:00

Yes.

Dominic: 15:02

We're a very innovative culture, actually, across our whole group. We have a central AI hub. We have innovation communities in our subsidiary here. And then there's functions like myself who provide enablement to those IT and data space. But before I give more details on that, the key I think for our culture is that everyone's an innovator.

Dominic: 15:27

We don't have this sort of gate kept, well no, no, no, don't do innovation, we do that. There is encouragement to be curious, to explore new things coming from our underwriting teams, operations teams, our HR marketing and within IT itself. So yeah, there is innovation everywhere. In terms of sort of business innovation, that tends to be those functions, those groups are there to support those ideas. When it comes to technology and innovation, our entire IT department is always looking at how we can improve our business outcomes.

Dominic: 15:57

So that's fostered and supported by our CIO. Very, very strong culture there.

Georgie: 16:02

How do you keep everyone up to date with all the innovative work that you're doing? So for a bit of context, we were at the Lloyd's coffee cart this morning. And somebody who works at one of our customers came up and I was explaining what we do. And she just went, I hate technology. I was like, Okay, give me that free coffee bag.

Georgie: 16:21

But she was just flat out like, I hate technology. My job has always been this way. And it's almost like they weren't aware of how much technology or how much digitalisation actually goes on in their organisation. It was maybe like a blanket approach is what she was saying. Because I was thinking, you probably use however many different systems to do your job.

Georgie: 16:49

And when I'm talking about what we do, it's not replacing that relationship human driven part, but it's how can I get you there quicker? And she was saying, I've made my career on having those relationships. And I was like, that is absolutely fair. I'm not here to replace that. What I'm here to do is help you not re key across multiple different systems, help you not have to chase people via email loads of times and enable you to access more markets, stuff like that.

Georgie: 17:18

And maybe it's that their culture doesn't share wide enough or celebrate the success that they're doing with technology and innovation. It's more gatekeep. So how do you share your wins across the business?

Dominic: 17:30

Communicate. I don't think that the adage of if you build it, they will come applies for a lot of these things. You have to build it, then you have to communicate. You have to really espouse the value this brings. But it's a really interesting story you gave there.

Dominic: 17:46

So there's a couple of things I want to draw from that. To me, when it comes to AI especially, there's a triangle of perception. And the one vertex, have people who believe it's gonna kill us all. It's gonna take over the world and we're done for. You have a group of people who believe that it's useless.

Dominic: 18:03

It won't really do anything. It's hype. And you just got FOMO. And then there's another vertex who believe that it's the best thing ever. It's going to solve all of our problems.

Dominic: 18:14

And we're all just going to sit back and relax because it's brilliant. Now in reality today, the truth is somewhere in the middle. Somewhere in the world. It's going to change our lives significantly. It will change jobs.

Dominic: 18:28

It can offer significant value in certain use cases, but it's quite limited in other use cases. We've all had conversations with ChatGPT where we've gone, oh, that's a bit stupid, isn't it? And then other times, it's got really intelligent insight or it can process data very quickly. There's a lot of myth there and the truth is somewhere in the middle and that will change. I'm sure the dial will move as it progresses.

Dominic: 18:49

What really struck me about your story is this idea of identity. If you went up to a baker and said, I'm going to automate your entire flow, can just sit back, just press a button, bake your bread. They'd say, well, I'm a baker. What does that mean for me? I'm just a button pusher.

Dominic: 19:08

And what you've got there is you've got human in the loop with pressing the button, but I hate that phrase as well because we should be looking at human in the driving seat. We should be looking at the baker coming with their expertise, their ability to mix ingredients in a particular way that works for them in their altitude, their environment, their humidity. And they can automate the bits that are a bit exhausting like kneading the bread. Or not going to stand there with the bellows, they're to put it in the oven. That's where the automation comes in.

Dominic: 19:35

That's where AI can speed up your life. And exactly as you described, it can open up new markets for this person. But their brain, their identity is still there. They're still running the show. But yeah, sorry, to answer your question, it's all about communication.

Dominic: 19:47

You've got to keep talking to people because the risk is twofold. You're either in a situation where people are dismissive and they go, well, it doesn't do anything, I'm not interested. Or people will go off and do it themselves and just duplicate what you've already done. And then you've got two, or you're spending time trying to reconcile those communicate, communicate, communicate.

Georgie: 20:08

Does that come back to you? So multiple different underwriting teams are wanting to do the same thing. You're like, Oh, alarm bells, we've already solved that over here. So how does that feed through?

Dominic: 20:17

Luckily for me, it doesn't just come to me. That's a bottleneck we'd want to avoid. We have a really strong solution architect community, very holistic view. As I said, we've got our innovation hubs. We have our data teams, we all collaborate and communicate all the time.

Dominic: 20:32

So it's very rare that someone won't have heard about something, oh, I think Dom's doing that or Dom thinks someone else is doing that, let's go have a conversation.

Georgie: 20:42

And last time when we met to discuss the podcast, you mentioned quantum computing. I'll be honest, I had to go and fully research what is quantum computing and fully digest what it is that we were talking about. So for the listeners, would you be able to explain what quantum computing actually is in simple terms?

Dominic: 21:01

In simple terms? Yes. No. I think if anyone came to you and said, I can explain quantum physics or quantum computing to you in fifteen minutes, they're lying. It's all gonna be simplified.

Dominic: 21:12

I'll do my best. Essentially, with computers, most people are familiar that you have bits of information, which are either one or zero, on or off. And a combination of those bits gives you the information, all the instructions that you're interested in. When it comes to quantum systems, you can have photons or other quantum particles, which can be one, zero, or a superposition of both. Now, going into too much detail, and I'm by no means the expert here, what that means is there are certain problems in classical computing which are very, very complicated and very long to solve, which quantum computing can shortcut.

Dominic: 21:51

And that's it, really. It's an optimisation approach using the fundamental laws of quantum physics.

Georgie: 21:56

So when we first spoke about it, I went away and did some research. And then I was in short tech week and went to ITI. And there was a panel discussion, and quantum computing was brought up. So is this the next hype that we're facing? Are we too early to start thinking about it?

Dominic: 22:14

No, definitely not too early. In fact, I wouldn't even look at this as a sequential thing. It's AI, let's wait till that's done, it's quantum. Look at them both. It's not at the point now where it's sort of enterprise scalable.

Dominic: 22:27

But you don't want to be in a position, just like the innovator's dilemma describes, where you're left behind and suddenly you're running to catch up. I would advise any organisation of any size really to start thinking about it and looking into it.

Georgie: 22:41

And in simple terms, is quantum computing going to replace AI or is it just that deeper level that goes further?

Dominic: 22:49

It may well, I don't know really, it may well augment AI. But AI is using these large language models, statistical predictive analysis. Quantum computing is a mechanism, a physical mechanism for implementing speed, really. These algorithms that can run on a quantum computer can be much faster. Perhaps AI can be much faster because of that.

Dominic: 23:09

But they're not replacing each other.

Georgie: 23:11

Where do you think the biggest potential impact of quantum computing could be in insurance?

Dominic: 23:16

I'd say there's two. One, both of them actually are applicable well beyond insurance. The first one that everyone should be thinking about today is security. One of the fundamental and early insights into what can be achieved with quantum computing is the ability to factorise very large prime numbers much faster than we can today. What that means is traditional security protocols are weak against quantum computing.

Georgie: 23:44

Ok

Dominic: 23:45

And the concern that people should have is this idea of harvest now, decrypt data. Essentially, you can take the data now, you can get access to it. You can take the data now, wait for quantum computing to be delivered at scale and then use those algorithms to decrypt that data. So protect your data, obviously. Everyone should be doing that anyway.

Dominic: 24:05

And prepare. There are algorithms provided now by and the big players are all implementing these that are written using classical computing concepts. But mathematically, they're safe against quantum decryption. So they're called quantum safe encryption rather than quantum encryption itself. So that's a big focus for everyone, not for insurance, for the entire industry.

Dominic: 24:28

It's very important.

Georgie: 24:29

Is that a potential new product line? Protection against quantum computing?

Dominic: 24:33

Well, perhaps. Yeah, good point. The other thing to think about is an opportunity. As I say, it's very efficient. There's an algorithm called Grover's algorithm, which essentially takes the efficiency of searching for something.

Dominic: 24:48

In the worst case scenario, you're looking for a red ball in a bag of blue balls. The worst case scenario, you have to check every ball because you're just really unlucky and you found the red ball as your last check. With Grover's algorithm, you can reduce that to the square root of the number of balls using, again, these superposition principles. And that's very generalist but it can be applied to any sort of problem that requires searching for a solution. So it's an optimisation approach.

Dominic: 25:18

So any complicated process that you could run-in insurance can be potentially sped up by the advent of quantum computing.

Georgie: 25:26

I feel like we have found our next hype. We're going to move beyond the threat of AI. That's just going to become everyday tool business as we use it. Then quantum computing is going to be the next threat that we have to face.

Dominic: 25:38

Yeah, and I don't even like the word threats. To me, it's these opportunities. We could have some real fun with this and we'll be able to find a business value there. That's really important.

Georgie: 25:49

Does bureaucracy actually hold organisations back? Or is it actually about how it is designed and used?

Dominic: 25:55

Absolutely, it is designed and used. Don't look at it as something that is dictating the way you operate. Look at it as an enabler for you to innovate, grow, explore, experiment safely. That's the key.

Georgie: 26:10

Yeah, I think the way that you've explained it is right. And it's the safe element. It's giving people the guardrails to go and experiment, but in a way that's not going to be a detriment to the business. And I think that's kind of it brings it all together. What have you seen work well when large organisations have successfully balanced governance with innovation?

Dominic: 26:29

Sandboxes. I had to come back to that sandbox word. Once you've produced that, once you've created an environment where essentially you can play with some safety nets, then you're gonna be able to unlock so much value, so much science, so much research that will deliver value to your business without causing safety problems.

Georgie: 26:53

And for leaders that are trying to move faster without losing control, where do you think that they should focus first?

Dominic: 26:58

Faster without losing control. Get your compliance teams, speak to the regulators, work with your data teams, data protection teams, Establish the foundations collaboratively. Don't sit in the dark room and figure it out yourself, communicate. Work with those teams. Build that environment in design mode before you even establish anything.

Dominic: 27:21

And you just accelerate like crazy.

Georgie: 27:23

How early do you think then those teams should be brought in when you're looking to do a pilot? Like should they be involved in the first stage conversation? Should they be involved when it's developed a bit more? Where should they be brought in for best practice?

Dominic: 27:35

So because they've been brought in right at the fundamental principles, some of their work is already done. They've established that environment there. Then it depends on your level of risk, what you're dealing with. If I'm just going to throw some fake documents into a chatbot and say, you summarise these? I want to prove a point.

Dominic: 27:56

As long as they're not company sensitive information, then you can safely do that on an informed basis. But at the moment where you say, well, I've got to put together this deep orchestrated AI solution with all of our policy information and this, that and the other, you want to take that a little more seriously and get those teams involved as soon as you start to design that experiment.

Georgie: 28:19

And so this is my final question. And this is, if we had a crystal ball, as you said, we would all be billionaires. But I'm going to ask you about your future outlook. So if we're looking ahead of AI, and I feel like I kind of already know your answer, but what do you think organisations need to prepare for beyond the hype of AI?

Dominic: 28:36

Beyond the hype of AI? Well, yeah, I mean, quantum, right? That might be a personal bias. Very interested in that there. AI is interesting itself, though.

Dominic: 28:45

We talk about artificial intelligence today. And this year is all about agentic AI, getting real value out of that. And there's a lot of talk about artificial general intelligence. Now I don't know where that's gonna go. I don't know if that's gonna go anywhere but it's something to keep an eye on because we're getting real deep human level intelligence if we should reach that there.

Dominic: 29:05

For now, I can't predict it. Keep an eye on it.

Georgie: 29:07

On the agentic point, where do you think people are at in terms of the trust with agents and the moving towards agent to agent communication. Because I think it's a staged approach. And again, this is another topic at ITI. Different stages. And they were saying stage four is agent to agent communication, which I don't believe in the specialty market anyone is there yet.

Georgie: 29:31

I think it's more having your own agent, as you were saying, that you could train, that you built in three days, and that's what people can ask information to and they basically become an encyclopedia for your company. But then the next level is us booking this podcast in today. You have an agent, I have an agent, and it knows our calendars and it does it for us. How far away do you think we are from that?

Dominic: 29:54

I'd say closer than you think.

Georgie: 29:56

But

Dominic: 29:59

there's important considerations there. What people should treat AI agents as is like employees. Let's imagine, this is total hypothetical, that you create an agentic AI solution which can pay a claim. Would you hire an intern and say, process these claims and no bounds, nothing? They come up to a £2,000,000 claim and they go, okay, boss said pay it and pay it.

Dominic: 30:26

You wouldn't do that. So you treat these AI agents as employees and you set them authority limits and thresholds just as you would a human. So you're not putting yourself in a situation where essentially you've just let things loose and you've caused huge problems to your organisation and then you can't even explain why because the agent did it. I don't know. And this is where we're going earlier with human in the loop and human in the driving seat.

Dominic: 30:49

The accountability rests with us. Today, you can't sue an AI agent. You can't hold that to account. You can turn it off. You can't really fire it.

Dominic: 31:00

So human in the driving seat, we use our brains. It's our accountability. But I think it's close. I think you'll start to see that across the industry with automated, orchestrated, agentic AI solutions. The world is getting interesting.

Georgie: 31:16

Well, I'd love to have you back on in a few years' time when we will be talking more all things quantum computing and probably have an agent here replacing. Yeah.

Dominic: 31:24

We want you to be here.

Georgie: 31:25

Yeah. Thank you so much for joining me.

Dominic: 31:27

Thank you.

Georgie: 31:29

Thank you so much for listening to Fact or Fiction. If you like what you hear, then please subscribe to our channels. We'd also love to hear from you about future guests so please get in touch via the link in the show notes or reach out via LinkedIn. Thanks so much again and we'll see you next time.