🦄 Making great presentations more accessible.
This project enhances multilingual accessibility and discoverability while preserving the original content. Detailed transcriptions and keyframes capture the nuances and technical insights that convey the full value of each session.
Note: A comprehensive list of re:Invent 2025 transcribed articles is available in this Spreadsheet!
Overview
📖 AWS re:Invent 2025 - AI, Cloud & Public Sector Transformation: PwC & AWS Driving Change Now (AIM115)
In this video, Catherine, Andy from PwC UK, and Nicola from NHS Medway Hospital discuss their successful implementation of AWS solutions including Amazon Connect, Lex, Lambda, and Bedrock to transform healthcare contact centers. Nicola shares how they reduced DNA rates by 4%, automated 80% of 110,000 annual calls within six weeks, and saved tens of thousands of patient minutes monthly. Andy highlights lessons from local government deployments, achieving 55% call automation in Birmingham with 80% customer satisfaction. The solution provides 24/7 multilingual support for diverse patient populations. They discuss challenges like understanding patient intent and clinical language, emphasizing the importance of staff engagement and iterative improvement. Future plans include expanding to cancer services, prison healthcare, HR/finance automation, and preoperative questionnaires, with ambitions for voice-to-voice AI and agentic interfaces.
; This article is entirely auto-generated while preserving the original presentation content as much as possible. Please note that there may be typos or inaccuracies.
Main Part
Transforming UK Public Services: Lessons from Local Government Success with Amazon Connect
Good morning everyone. Thank you for joining us this morning. I'm Catherine, a director in the UK PwC firm working in the health team. I'm joined today with my colleague Andy, who works in our local public sector team in the UK firm, and Nicola, our client who works for the National Health Service in the UK at a hospital just outside of London. We want to come and talk to you today about how amazing AWS products have been transforming the public sector and the work that we're doing in the UK.
I think for those of you who don't really know much about the UK, the National Health Service is actually the biggest employer in Europe. Despite the fact that you've got millions of people working for the NHS, the waiting list for access to healthcare is 10% of the population on a waiting list. They just don't have enough capacity to see all of the people, all of the customers who want to get access to health. And the fact that it's the most trusted institution in the UK, the waiting times are basically what dominate the citizen perception, giving it a really bad reputation.
Even though the NHS app is actually one of the most used healthcare apps in the world, the technology landscape in the NHS in the UK is pretty complex, really outdated, and has very limited use of digital, AI, and cloud. This actually gives us a huge opportunity though to do some really exciting stuff to improve that citizen experience. So Andy, we obviously learned a huge amount from the work that you were doing in local councils. Do you want to reflect on the learnings that you've got from that work?
Yeah, sure. So I think as you'd have seen from the video, we started off working with Amazon Connect in local councils, so local government in the UK, and I think we found three things that we can do really quickly. The first thing was that you can do things really fast, so we can deploy Amazon Connect in 10 to 15 days for a full contact center migration. We can release 30% of a typical contact center's capacity in a matter of 10 weeks. And with the right governance and decision making, that means you can achieve results really fast. One of the things we need to do is take that learning and transfer it into the health service.
The second thing is there's a lot of components, and once you've created once, you've got a reusable asset. That builds in quality, that builds in pace, that builds in a lot of things that mean that we don't have to worry so much about the specific requirements of a given industry. We can move from one to the next very, very fast. And the last thing is councils really do need to modernize fast. They're running out of money, they have increasing citizen expectations. The two components that we put together meant that people were willing to say, all right, I'm going to go and try that. I know I can fail fast, I know I'm not committed if it doesn't work, let's get involved and let's get it done quickly.
Addressing NHS Medway's Contact Center Crisis: 50% Unanswered Calls and Patient Access Challenges
Amazing. And so I think having heard that story and the success that we had in local government, Nicola, what was it that really interested you and how we could apply it into the NHS? Yeah, so I think as you said, patients are waiting easily a year for a routine appointment. So when I say we have a long wait, we have a long wait. On top of that, Medway where I work is an incredibly deprived area. We have really poor engagement with healthcare, huge numbers of drug abuse, alcohol abuse, social deprivation, and it makes it harder for our patients to be able to access our healthcare.
If I tell you that our average reading age of our demographic is six, so incredibly challenged. One of the things that we really wanted to understand is how we can improve how we communicate with our patients and how our patients are able to communicate with us. Not only that, by starting work with PwC at the very beginning, we average at around 120,000 calls into our contact center a year. 50% of those were going unanswered, and that's not because our staff were sat there doing nothing. It was the sheer volume of numbers that converts into patients that either want to change their appointment, can't attend their appointment, and therefore we don't know about that, we have wasted appointments.
Therefore you're waiting even longer than a year because we're not utilizing our opportunity within our clinic templates. So we really wanted to understand how we could work with a company like AWS to try and put something in that really supported our patients. But not only that, it supported our staff. Because we weren't answering the calls, we were getting a lot of very angry patients. And that was coming down onto our staff that are some of our lower paid staff and were coming in every day to a barrage of abuse from patients, because they've waited for a year for their appointment, and they've tried to call 12 times and no one's answering the call.
I think we were looking at it not from the perspective of could we have AI and what can it fix, but rather where are our problems and how can AI potentially support some of the issues that we've got as an organization. It's fairly easy if you look across what citizens want and what they're calling up for. The majority of that, about 80%, is really easily self-served. I think the majority of them was where do I park my car, which is a really low bar for citizen experience.
Andy, I think you mentioned earlier about the fact that it's quite financially challenged, and I think everything is around outcomes and how you improve that. What is it you in particular found around improving customer experience whilst also reducing cost? I think there's probably four things to call out. The first is that using Lex and Lambda, we're able to automate a large volume of that simple transactional contact. That means that calls never go unanswered, it means that there's an effective answer each time, and it means that that's capacity that's not spent on the phone when it could be doing something more valuable.
Certainly what we saw in Birmingham, the example that you probably saw on the video there, was that 55% of inbound calls could be automated and have been automated. That's 55% of total volumes of calls attempted, not just those that an agent could get to. The second is that we can drive really high levels of quality, so 80% of all calls in Birmingham again were answered effectively to the customer's satisfaction and with that journey being fulfilled. So we've now got high levels of automation delivering a high level of quality.
The other is it's available 24/7 because the robot doesn't sleep, and that gives us a huge advantage for some of those marginalized communities who perhaps can't contact in the hours that traditionally public sector organizations would offer. But the third thing, and I think the thing that makes it really powerful, is that by connecting again with Lambda into our back office systems, we can actually fulfill the end-to-end journey. So you want to raise the case? Yeah, we can do that. You want to start and process an eligibility assessment? Yeah, we can do that.
It means that we weren't providing an automation to just hold someone in stasis. We were actually delivering what they needed when they needed it accurately, first time right every time. Fantastic, and I think Nicola, it really resonates when you say this isn't just about AI as a buzzword or where can we throw it at. It's actually really understanding what problems you're trying to solve. So obviously we implemented Connect, Bedrock, Lex, and Lambda. What in particular was it that you were trying to solve and what was the impact that this technology had?
Rapid Implementation and Results: 80% Call Automation Achieved in Six Weeks
I think Andy's referenced it beautifully with how transferable this kind of automation is. As I said, we wanted to start it with a basic rollout within the organization. I'll apologize now if any of you are clinicians in the room. They are the hardest group of people to work with when trying to put any kind of change in. So we didn't want to come full bang with it and say let's automate radiology reporting. Let's take it away from that and actually understand what we can do so it's scalable and we can then build up to doing those things.
I think for us with our contact center, we wanted to understand the number of patients that were now able to get through. So we have something called a DNA, which is a did not attend. We monitor that on a percentage on a weekly basis. We can understand how many of our patients didn't come for their appointment and why they did not come. That has dropped by 4% since we started this, which doesn't sound huge, but it's a massive amount of patients for us.
Within six weeks, it took six weeks to deploy this within our area once we had the go-ahead and got through all the information governance element of all of it. 80% of our calls are now automated, which means that out of 110,000, that's 80,000 to 90,000 a year that are not having to have a person answer that. We've always given the patients the ability to have a human being there. We have a lot of elderly patients, a lot of patients that have neurodiversity and have learning difficulties. There's still that human person if they want to talk to them.
But we've saved, I think, patient minutes. We're saving tens of thousands a month in the fact that those calls are answered straight away now. As you referenced with the council, it's open 24/7. It's not just open 9 to 5 while people are working normally. It does it in multiple languages. We have a huge immigrant population that are non-English speaking as their first language, so again that accessibility that is given to our patients has gone through the roof.
The patient feedback we're receiving now has gone from that cohort of patients that, as I said, were giving our staff abuse. It's flipped completely to now an incredibly happy group of patients that are saying why did we not do this years and years ago.
I think the impact is massively tangible for us. It's really exciting and such a compelling story for something that's pretty simple.
Obviously not everything is a really easy ride. Andy, what were the kind of challenges that you in particular faced and, more importantly, what are the specifics that are really important to get right when you're deploying this? One of the things we take for granted is that the labels we give to things mean anything to our customers and our citizens. Actually, the title of something that's used internally may have no meaning whatsoever to your average person on the street who just wants a thing. So the first thing you've got to get your head around is how do we accurately understand their intent. It's not about what are the words they're using, it's what is the thing they're trying to do and how we're using the technology to help us understand that and resolve it down to a thing we can then act on.
There was an interesting example in Medway where there's obviously some clinical language that's used, which is beyond the sort of normal parameters that something like Lex would be able to accommodate. So then there's a bedrock lookup that can go and find the really complex medical language and resolve that down so it's accurate because that's one thing you don't want to get wrong. The second thing is you've probably heard that phrase about don't let perfect be the enemy of good, and coming back to clinicians being harsh judges, you'll never get something that's perfect, and you'll never get something that's 100% right. But half of the job here is to get your MVP at 80%, put that into production, and use the feedback that you get to make it as close to perfect as you can.
But of course what that relies on is that you've got the ability to change things quickly. So again, this is where the tools set really comes into its own, because you can make those changes as you go, as part of the continuous improvement method of the actual service that you're engaging with. It's that fusion of, if you like, the business process and the technology configuration that means you can make progress very, very quickly.
Scaling Beyond Contact Centers: Future Vision for Clinical Services and Agentic AI
That's great, and I think the work that was done at Medway is so replicable that we're now putting it into multiple other NHS organizations. I think the learnings from that of what you can get done in six weeks, and Nicola, that's a phenomenally rapid period of time to get this up and running. Where did you kind of take it next and what's the kind of future for you?
I'm going to start by saying the fact that I am stood at this conference talking to you guys, I am the least techie person you are ever going to meet. The fact I know what agentic AI now is, I have sat there listening to the geniuses at AWS talking, Googling stuff as they're talking to me. This is so exciting for me and for my trust. We are not reputationally known as being a transformational organization. The fact that we have big London teaching hospitals coming to us and asking us how we've done this, how we've done this so quickly, and I think a lot of that has been around the engagement, how we've engaged with our staff.
If you look at our contact center, we didn't go in because the minute you talk about AI people think well that's it, my job's gone, I'm losing my job, they're replacing me with a computer. So we've been really careful about how we've gone to those teams and involved them right from the beginning. They've been part of the design and I think the reason that we did that is because we wanted them to support what we're doing, because if they weren't going to do that, this was never going to work, and I think that's huge.
Next for us, oh God, where do we go? It's the sky's the limit. The scalability of this now for us as an organization, we've done a number, whilst we've been deploying, we've been doing scoping work to understand what can we do. If I give you an example, within Medway we have five prisons. So just for us, our prison population is huge. They are again a massively challenged demographic of patients to reach out to and get that engagement with. So we're now looking at how we get a regional hub within Kent and Medway to support with all of that for other hospital trusts within our own localities.
We're rolling out the work we've already done to other areas, so we have another where cancer is the next one we're going to that's built and ready to be deployed as we speak. But we're also looking, the NHS is under a massive amount of financial pressure, huge amount. We have huge efficiencies. My organization has a 50 million pounds saving target this year which I know in AWS terms that's pocket change, but for an NHS hospital trust it's a huge amount of money and we have to look at that awful how do we take people out, which is not a nice thing to do.
But we're now looking at what we can roll out in some of our back office functions, so looking at HR finance. Again those, it's the same methodology, but do you need someone to tell you how much annual leave you've got left or when your payslip should be coming out. So again it's looking at where we can scale this to. And then I think the next move, which is going to be incredibly exciting, is how we start to bring in more clinical services.
For example, preoperative questionnaires don't need someone to be manually administering or reviewing them. You've got your set scripts, and depending on the answers, patients will go into a green box indicating they're good to go for their surgery, or they'll be flagged as needing a review. Again, it's taking the pressure away from our clinical staff but also from our administrative staff. I think every time we start to scope, we see more opportunity, and it's so exciting. I don't think I ever thought working in an organization like the NHS would give me the opportunity to do this. I flew over from London yesterday to do this, which is amazing. So yeah, I think the sky really is the limit for us.
Yeah, and I think that's so exciting, especially the mindset in this particular instance that made it so successful. Just really willing to give something a go, leverage what has been built successfully in other places, and keep scaling it and rolling it out. It sounds like a really ambitious vision for what you want to create across your geographical area. But I guess, Andy, just closing off with you, what's in your sort of ambitious future?
Yeah, so one of the things that using Lex to do this encourages or invites your customer to do is perhaps sacrifice a little bit of the customer experience for a lot of capacity release. But with Nova Sonic arriving and a range of models that are going to do voice to voice, we can actually redress that balance. So we're making a big investment right now about saying how do we take everything we've learned and put it in that purely generative environment so we've got rich conversations, high customer experience, empathy, and intimacy perhaps where it's appropriate to drive greater demand for the voice channel at very, very little cost.
The second thing, picking up a point Nicola made, is everything that your customer wants from you as an organization, your staff will want from you in some other way, some transactional need or requirement. There's no reason why you can't automate business support using the same tools that you use for customer support, and I think that becomes quite an interesting two-dimensional approach to using Amazon Connect.
Then the third one, and I think this is where it gets really exciting, and you've heard lots of announcements already about the sort of agentic developments, is why do I need my citizen or customer to phone me up? Why can't I just get them to say my agent's going to deal with that, and we'll create an interface that allows their personal agent to achieve all the things with us as the council or the hospital that we need without having to lift a finger or a microphone? I think what becomes quite exciting about that is we can really empower people on their behalf to get the best things proactively rather than waiting for them to contact us. So I think watch this space. There's a lot coming there.
Fantastic. Great. Thank you very much. Thank you. Thank you all.
; This article is entirely auto-generated using Amazon Bedrock.
Top comments (0)