Transcript
Speaker: I will be the first to say AI is not the answer to everything. Sometimes you need judgment and you need human judgment to do the most critical decisioning task.
Speaker: Hello and welcome to Edison TV. Today I'm joined by Samir Bhujuru, Chief AI and Product Officer at Capita. Capita are a leading UK business business outsourcing company with a particular focus on delivering solutions for highly regulated and complex environments, particularly to the the UK government sector.
Speaker: The company's in the middle of a substantial turnaround, really with artificial intelligence underpinning that and and Samir is central to that transformation. transformation Samir, many thanks for joining us today.
Speaker: Can you start by introducing yourself? you You joined Capita from AWS, a senior role at AWS. Can you explain why you made that move and and why in particular you believe that Capita is poised to benefit from artificial intelligence and the difference you can make at the business?
Speaker: It's been about 18 months since I moved from AWS and at amazon Amazon Web Services I was serving the global telco industry. And when this wave of AI was unleashed, we rightly went to our customers and we said, look at what this technology can do.
Speaker: Isn't this amazing? Give us your most complex operating procedures, and we will take them off your plate. We'll simplify them, and we'll just hand you something that does it for you.
Speaker: So we went in to our customers, and they handed us a standard operating procedure document. The tech wasn't the problem. We would agentify that in days, go back and test it in real life, and it would fail.
Speaker: And when we diagnosed why it would fail, it was always because real-life process execution bore very little resemblance to what was in that standard operating procedure.
Speaker: And when you follow that thread, the people who owned and were executing that ah standard operating procedure in real life were the business process outsourcers. So you had to work backwards from what worked from the nature of work itself as opposed to the idealized nature of work which was captured in those documents that we were identifying.
Speaker: And so I saw an opportunity to be at ground zero for agentification because if you believe where the value is moving, human labor is going to be significantly augmented and the productivity is going to improve 100x to 500x depending on which market researcher you believe.
Speaker: But you can't argue with the quantum. the And it is significantly bigger than the cloud or the internet. maybe even bigger than the industrial revolution, and look at the productivity that unleashed.
Speaker: And we believe business process outsourcers have a significant role to play, if not the most significant role in adopting AI agents.
Speaker: Thank you. and and And looking at, I guess, a little bit more detail in terms of um how Capita is deploying AI, how is it how how is improving the company's operating performance and competitiveness now? and And how do you see that evolving going forward?
Speaker: So there's three ways that I really think of AI in our operations and how that translates into value creation. First one is it significantly reduces our cost to serve.
Speaker: Historically, what used to take years to implement. with the advent of AI-assisted coding and all of the artifacts made available by the frontier firms, we can drastically shrink that timeline and the cost to actually deliver something to our customers.
Speaker: And that cost to serve translates into a real advantage when we are bidding for new business. So you should see that, and we are starting to see that, as a significant improvement in win rates.
Speaker: So that's vector number one. Vector number two is we run a lot of complex processes and we run it through shared services.
Speaker: We run payroll, we answer calls, we run ah transport for London's road user charging, we run BBC TV licensing and all of those are incredibly complex processes with tens if not hundreds of steps and some of those are incredibly boring and pure drudgery.
Speaker: So actually taking away the the the constrictions in that pipeline opens up productivity, which means we can do more. with the same number of people, which means that ah that becomes a productivity-driven margin improvement.
Speaker: And then the third one is actual agents out there doing what used to be done historically by an army of people, typically in an offshore location.
Speaker: But this gives us an opportunity to have highly skilled people operating that technology onshore for significantly less cost than offshoring.
Speaker: So those are the three ways that you should and we're starting to see AI play out into monetization. So interesting, so value moving moving towards business processes rather than technology implementation, if you like. um Looking at Capita in particular, you're you're very and strong and focused on government public services.
Speaker: How does the implementation of artificial intelligence um differ from other environments where regulation, compliance and and elements like that are so much more important?
Speaker: So if you look at the public sector, and and let's just take the UK because that is the majority of our business today. We require local citizens, so it is an employability and an employment uplift for the UK, whereas historically you may have outsourced them to low cost geographical locations.
Speaker: It is a total upscaling of not just our workforce, but everyone who is using AI. And it is democratizing that AI access to the the government departments that need it.
Speaker: And we'll be one of those players, we hope, that does that. And finally, when you operate on highly sensitive information, like like the u UK government holds in multiple departments, you want security clear personnel who are really cognizant and capable of operating the AI.
Speaker: So we we we're really invested in getting our customers' AI ready. And then on the other end, it's also about the concept of forward deployed orchestrators.
Speaker: Because you can implement a use case with the tools that are available right now, but that use case will drift over time. as the models change, as costs change almost daily, apparently, these days, as regulations change, as the needs and scale of the people we're serving change, and how they expect to consume these services change.
Speaker: So you need an AI capable security cleared staff that can operate this AI in real life for the decade or decades that this particular process needs to be run.
Speaker: And that's where we come in with help. And can you discuss where Capita sits within the ecosystem? If you look at the overall ah AI delivery stack, if you like, you've got hyperscalers, you've got the systems integrators, the software providers and and and specialist specialist AI companies. um yeah Where does Capita sit and in in that ecosystem? We are none of those, but we are an incredibly important partner to all of them, is the way I would describe it.
Speaker: Because at the end of the day, our reason for existence and what we do really, really well is we deliver outcomes using complex processes at scale.
Speaker: And that understanding of what can be done, how was it done historically, how can it be done better in an agentic first world, and how do you actually translate that into something that the firms that you just talked about can actually execute Think of it as getting the process AI ready.
Speaker: Not the infrastructure AI ready, not the models, none of that. Let's get the process AI ready and then use the fantastic tools that we all have at our disposal today to get that AI ready process actually agentified and helping our frontline workers do their job better.
Speaker: ah Can you talk a little bit more depth about how you managed to integrate AI into these complex environments, these complex complex workflows that you were discussing, where regulation, orchestration and and so on are much more complex?
Speaker: First of all, um I will be the first to say AI is not the answer to everything. Sometimes you need judgment and you need human judgment to do the most critical decisioning tasks.
Speaker: So you have to identify that upfront. Which parts can be automated? Which parts can you you apply AI to? Because remember, AI will learn incredibly fast, but in the beginning, it is not deterministic.
Speaker: So you do not want to deploy AI where you need a deterministic outcome in the beginning. You want to deploy it side by side with a human and with old school automation and traditional AI, which is more deterministic.
Speaker: And so understanding the stages of adoption and getting to that truly intelligent AI deployment is what we do really well, I believe.
Speaker: And where the industry struggles sometimes is they try to get to that end state and say, AI is the answer to everything. AI will be the answer to everything, but you've got to go through these phases.
Speaker: You've got treat the human element humanely. You've got to treat the automation element, the traditional AI, with the respect that it deserves because it's been an operational and delivering results for decades now.
Speaker: And then you will get to the promised land of fully agentified business processes. But it's a journey. And then and then looking um at where AI has been deployed in in in the current environment, where where are you seeing the strongest demand um for, I guess, AI-enabled services? and And then looking forward, where do you think the big opportunities are for Capita?
Speaker: So I'll take a couple of examples. The front office, which is when citizens contact us, there there is incredible adoption and acceptance that AI is going to play a role there.
Speaker: And we are deploying AI across our internal operations to assess the frontline workers and to make sure that they have a better quality of employment, because that then translates into better quality of service for our citizens.
Speaker: And so that that is what we're focused on, our internal operations, making sure that they are optimal in the front office. And an example of that would be out-of-hours support.
Speaker: No one wants the 2 a.m. to 6 a.m. shift on Friday night or Saturday night. If you were to intelligently deploy AI during those hours, you'd have a lot less but better paid people because of overtime when there is an escalation required.
Speaker: So that's an obvious one, and that it is progressing at scale today in the UK, in the public sector. But where we see the real value is in the back office in the middle of and the middle office. The back office, for those of you who don't know, is essentially all of the asynchronous transactions, like auditing, ah as an example.
Speaker: Whereas the middle office is an actual case flow. Whereas you need something from your government or from your council and you basically ask for it and then it goes through a case flow. We all do that almost on a weekly, monthly basis.
Speaker: And so automating that is our singular focus today because that throughput, when we're able to achieve it, drives down waiting times. Again, better citizen satisfaction, better public services.
Speaker: Thank you. And as we've discussed artificial intelligence, you're underpinned with deep human expertise is fundamental to your transformation program. um what should we know you know How should we calibrate that? yeah know How fundamental and how big a difference do you think the deployment of AI can make to your margins, to your customer relationships, to your your compliance and so on?
Speaker: like like like Like any good company, we have targets internally that we believe that AI will deliver. And they are pretty significant, Dan. I'm not at liberty to say exactly what the margin expansion targets. We've set ourselves up, but they're incredibly aggressive.
Speaker: And ah you should see a material improvement if we were successful in our in our strategy. And you should see that in increased win rate.
Speaker: You should see that in lower cost to serve, aka lower cost of goods. And those are the two metrics I'd be watching out for. Samir, many thanks for coming in. Fascinating discussion and look forward to following the progress as as you move along the pathway.
Speaker: I'm looking forward to talking to you again, Dad, when we've made more progress. Perfect. Thank you so much. Thanks, Samir.
