Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Deep Learning with PolyAI

Deep Learning with PolyAI artwork

What if your contact center became a product team?

Nov 17, 202635 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Three years ago, Simplyhealth's PolyAI deployment was only resolving simple FAQs over the phone. Today, they run voice, in-app, email, chat, and WhatsApp conversations on the platform. Nikola Mrkšić talks with Paul Atkins, Head of Business Optimisation at Simplyhealth, about retraining a service team into AI builders, how they fixed a repeat-contact problem, and a hackathon that shipped a working agent in one afternoon. Listen to the full episode, and learn more about PolyAI at https://poly.ai?utm_source=youtube&utm_medium=podcast&utm_campaign=podcast&utm_content=podcast #PolyAI #VoiceAI #CustomerExperience #Poly Poly.ai Follow PolyAI on LinkedIn Watch this and other episodes of the Deep Learning pod on YouTube

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_00You don't need to tell your advisors that they need to become software engineers, right? You can start with like adjacent capabilities they already have, like conversation design, knowledge, QA, customer insight, root cause analysis, journey improvement, and then add the technology skills around those strengths and allow them the freedom and the tooling to be able to try, experiment, learn, and support them in that way. And as we, you know, as we've learned ourselves and we've learned in our leadership team here at Simply Health that the results can be incredible.

SPEAKER_01Hi everyone. Welcome to another episode of Deep Learning with Polyai. In this one, we have tried to help skip leaders get a window into the latest and greatest developments in AI and automation. Before we start, reminded to you know like, share, and subscribe. And today I have the privilege of hosting Paul Atkins, who is the head of business optimization at Simply Help, a long-standing client of ours. And he spent the last three years building one of the most impressive dialogue AI deployments that we've ever had.

SPEAKER_01You know, I think we started off on a very transactional use case basis. Now it's grown into a great partnership into something much bigger and to change how teams run. So I kind of maybe want to start there and then we can work through how it all happens. So, Paul, thank you. Thanks for being here. It's great to have you on.

SPEAKER_00Thank you very much. Hello, everyone. Yes, it's been a great journey, right? I think we started oh 2023, I want to say. March 2023, something like that. And partnership with Poly AI really started in the voice channel, as as many have and will. And I think we uh started really small with some kind of transactional use cases around resolving AQs and that kind of thing. And I think since post-launch, we quite quickly got into building out those FAQs, understanding how it worked, and more importantly, understanding customer journeys.

SPEAKER_00And we were into resolving around 800 calls a week, which was a decent chunk of what we had. And then since then, yeah, we've we've kind of evolved to where we are now, which is which is using all the things you have and helping helping you helping you help us dive into new channels like in app and email responses and old chat as well. So we are, yes, we are. We've had quite the journey. So thank you for having me.

SPEAKER_01Yeah. No, no, absolutely. It's great, it's great to have you on. And look, I mean, I think that like what's exciting here as you kind of like start, you know, with like a small thing and then it builds and builds while you see the whole technology stack get rewritten and then like new tools come in. You know, I think that the traditional contact center, and even like the modern one, is a difficult place, right? I think that you're kind of like a firefighter without a fire hose, you know. Like I what I've kind of learned at least is that for the most part, you're worrying about whether you're gonna have enough people, then if you somehow solve that problem, then you're worried about whether you've rammed them up and onboarded them properly, right?

SPEAKER_01And only then do you maybe get to kind of like, you know, look at the things they're using. And you know, the most urgent things would kind of like get sorted out as you go, but like everything else was just kind of like left in a perpetual status quo. Whereas I kind of feel like now as we like bring AI labor into the whole thing to do much of the like tier one work, it finally allows people like you to step back and kind of go, like, all right, what uh what are we really doing? There is observability, there's just the chance to actually revamp the process to be a bit more of a product team, right?

SPEAKER_01And like in theory, it was always supposed to be that, but I think it was just you know, you would spend most of the time ensuring that there is obedience with your process that probably wasn't even optimal to begin with. And you would rely on humans to do the rational thing when it's not. But anyways, to tell me a bit about that. And you know, I've

SPEAKER_00yeah,

SPEAKER_01much already. I think I'm I think for the audience it's much more interesting to hear kind of like how you looked at it when you started and then like what the work looks like today.

SPEAKER_00Yeah, I think the real story for us isn't that we kind of installed a voice, but I think conversational AI has changed how we think about service, I would say. So we've moved from sort of buying automation to developing internal capabilities with people who can't analyze conversations, like redesign journeys, build knowledge, test changes, and progressively extend that intelligence across across channels ultimately. So we haven't really we haven't mainly viewed AI simply as like a headcount-based exercise.

SPEAKER_00What's changed is the nature of that work, I would say. So as more routine demands can be resolved through through kind of our AI solutions, our people increasingly are starting to focus on conversation design, knowledge, analytics, testing, root cause, a continuous improvement. So we're no longer sort of operating a service. Increasingly, what's happening in our organization is the people closest to the customer are helping to shape and improve the technology behind that service and dynamically shape our customer journeys.

SPEAKER_00And I think for an organization like Simply Health, which is um twice as old as the NHS in the UK, so it's 150 plus years old, and we we have a mission to help you know people in the UK get healthcare when they need it, as quickly as they need it, and support our NHS in doing that. I think kind of with that mission in mind, we've started to adopt more of a product ownership-based approach in which we partner with our product teams. Our products and services, from a digital perspective, are moving as quick as quick as they ever have been.

SPEAKER_00And what we're doing now is helping architect the experiences and design the experiences that sit behind that so we can keep pace with the with the product development that's happening over here, right? And I think that's been the real change and shift change for since we started the partnership with PolyAI. It's been from solving service use cases to now being an integral part of our product development life cycles, right? And evolution ultimately.

SPEAKER_01Yeah. And I think that with many tech companies, you see the same thing, especially in these like forward-deployed motions where you know you get this like kind of like flat from builders, and then you have like people who are actually like on the front lines doing it. And I think that like at least in customer service for the longest of times, it was so hard to keep the front line like well, to get it to hold, that it was just so hard to get any information back. And it's you know, kind of like the paradox of voice of the customer programs costing companies millions while you have, you know, hundreds, if not thousands, of people speaking with your customers every day, right?

SPEAKER_01So, like how do you see that kind of like gap being filled? Is it being filled? Like, do you feel like there's finally like good signal coming back from the contact center into the product?

SPEAKER_00Yes and no. I think it's I think it's improving as we understand this new world and what it can give us and how we deliver that information back into the product evolution lifecycle. I think there are still large gaps between staying in the transactional, being metric focused around resolution, containment, whatever you want to call it, and and service level agreements on a B2B side of things. I think there's still a huge focus transactionally and and in the service sense at that part of the experience that your organization offers.

SPEAKER_00And I think there is still a gap between kind of being able to step back from that and understand what a journey looks like and what the root causes upstream and downstream impacts are in that journey, and being able to then co-architect that, use those, uh build those epics, use cases and work with your product team to to almost fold the service wrap into the overall product uh design experience. I think there's still a gap between those two things occurring, but I think it's getting closer and closer. There are, you know, we work with uh Edge tier for a lot of our insights and analytics from a conversational perspective, and we use a lot of that data alongside what we get through analyzing our voice interactions with Polly and Ren to help build out those use cases and cases for change and transformation, but also to help us continually and dynamically improve the customer journeys that that support those those experiences that the you know the the interactions um give us feedback about.

SPEAKER_01Totally. I mean, like you mentioned, I think that like that's been one of the most exciting developments for me, and that at least maybe for the audience traditionally, you know, the business side is the business side and they own kind of like the procedures, the operating procedures, the source of truth. But then, you know, when you wanted to implement even like an old school DTMF IBR or anything else, you'd kind of go to IT and go like, well, I would like this, right?

SPEAKER_00Yes.

SPEAKER_01And I think that in the early days of conversation, and kind of like, you know, when we started our partnership, we were really still bridging that gap with forward deployed teams where we would kind of like come in with our like deployment strats and work with you almost as, you know, a mercenary army on loan to be your like second IT team. So that we almost circumvent the traditional enterprise, you know, kind of need for like, okay, we want this budget approved, next year work gets done. Oh, we didn't scope it right.

SPEAKER_01We need a bit more, we don't have enough budget. So that's I think where a lot of this was stuck historically. I think the forward deployment wasn't at least allowed for like a highly motivated group of people on our end to work hand in hand with you. And you know, for like large volumes like yours that made sense, right? But I think like, you know, one thing that I've recently been kind of like just repeating ad nauseum is when you have a strong partnership, like we had. Well, the working theory, if you know that the whole 10,000 hours to get good at just about anything.

SPEAKER_01My hypothesis is that a good forward-deployed team that is like very strongly aligned with the customer can probably do about 3,000 of them, right? And the first 3,000 are probably, you know, they get you half of the way there. And it can be very useful, right? It gets like decent resolution, it can be like a very productive application, right? But then kind of like as you really work to make it phenomenal, no amount of like motivated external third party will know, care, get in as deep with your business as you can.

SPEAKER_01And you know, we have like some behemoth deployments where people have worked for you know Fortune 50 companies for like three, four years. But you know, I joke with our champions there that they're basically like not working for us, they're working for them. That's fine. Because you know, if someone joins Pali out of university, they pretty much just worked for that company their whole life, right? But no matter what, they don't have the authority to then make some of those decisions, right? And they're not able to just kind of like connect inside the organization. So I think to do the full 10,000, you just have to put the tool in the hands of those that are empowered, that know how to get the data, cross-functional agreement of those things.

SPEAKER_01But uh how have you found that whole experience where you know we went from us doing most of the work to us doing all the snowwork?

SPEAKER_00Enlightening, right? I think it's been and it's been really interesting in terms of the dynamics of a of a partnership, right? It's not it's not been like your traditional deployment partnership that you might get with other other organizations, or you know, indeed a third party uh deployment partnership. I think it's it's been really interesting that kind of my team specifically actually have grown in terms of skill and knowledge and using the using the platform and software quite drastically over that time since we started when we heavily relied upon upon the teams at Polyai.

SPEAKER_00And I think now it's almost flipped on its head. We're coming to you guys saying, Hey, can we do this? Have you got something that can do this? You know, how do you think the product can flex to what we want to do? We want to look at an app, we want to look at email responses and stuff. How can we push the product to meet our business demands, but hopefully provide you some product insight as well, right? And some avenues for product evolution as well. So it's been quite symbiotic in that sense, I would say, in that there's been winners all around the camp, right? I think in terms of how we've managed to get great value out of the team, improve our skills using the software, solve real-time business cases on a on a large scale, right, deliver value back into the organization, as well as hopefully help the evolution of Poly as a product as well, and help inform that and give some real user experience of what these what the functionalities and various elements can bring in the real world.

SPEAKER_00And I think to your point on Ren, that's been a bit of a game changer for us, right? It's dramatically for us reduced the distance between customer conversation and actual insight. So things that you might have spent hours manually reviewing, and that's you know, that has been the case. People spend less time looking for things, less time building and checking things in the in the customer journey within the platform itself, and much more time now thinking about actually how we are strategically building these customer experiences ultimately.

SPEAKER_00A really good recent example, we combined REM with agent memory and asked it to analyse repeat contact for us. A question you're gonna get asked a million times by your operational leadership, right? Well, why am I getting repeat contact? What is it? Is it changing? And on day one, it identified that kind of 80 of the 550 conversations we had, around 14% were repeat callers. And the important thing was it didn't just give us that number, it was starting to tell us why those people were coming back in the context of their because we've got some good API integrations now, right?

SPEAKER_00So in the context of their experience, their customer lifecycle, what the products they have, and all that kind of stuff. And yeah, it was it was brilliant to be able to get that rather than having to put all the legwork into you know, mining that information out, guesswork, validate with qualit uh qualitative information, all that kind of stuff, right? It was it was there. And so the the life of a team like mine, which is transformation and support base, has been changed dramatically by this, by this tooling, right? It it allows us to be able to answer those things way, way quicker than we previously were.

SPEAKER_00So yeah, and and and now we're using uh obviously ren within the platform itself to, you know, as I said, I gave you an example earlier of uh how we kind of hacked together a we did a hackathon in in our technology team, and we spent an afternoon building out a new agent for sales support teams, and the whole team got involved, did it, looked at the FAQs, looked at the journey, looked at the systems, looked at the contact data, and then we used Ren and we built we built an agent and we're gonna deploy it next week. And we did that, you know, as a team, as a group bonding exercise, but also, you know, it served a practical purpose and it solved the business problem as well, right?

SPEAKER_00In the space of a few hours. So um, yeah, incredible.

SPEAKER_01I mean, I think like Ren has been just like an absolute, like, you know, a lightning in a teacup moment for us, where, you know, as we kind of like had like our previous tool, Smart Analyst, which was like a

unknowngood,

SPEAKER_01I think, analytical tool. And you know, it was used, but you know, it was just one part of the platform. There was a point where, you know, as coding agents started becoming sufficiently good that we allowed it to do more and more. And then at one point, I realized that with some of the beta customers that had it, there were a few people that were sending two, three thousand messages per month back and forth. And I think in that whole like technological leap, like the CX people became the builders because it stopped being about justifying the effort so that other people can go do the work, get the data back.

SPEAKER_01Because those things I feel were like if you were on a roll, if one thing worked, and maybe you got to ask for another and maybe another, but the cost of it, regardless, was quite high, right? You would have to go and just like do the work, listen to the calls. It was hard, right? And like listening to calls, whether it's closed between humans or between AI, my work theory is like if you can do 50 a day, you're like an absolute tier one hero, right? Like it's boring, it gets really boring, right? So no human being can do what AI can in terms of just surveying them all.

SPEAKER_01But tell me a bit about like that transition of the team from like service to tech. I'm not really sure it's service or tech. To me, it's just really that whole like

SPEAKER_00it's

SPEAKER_01combined effort, but

SPEAKER_00yeah,

SPEAKER_01yeah.

SPEAKER_00And the role, the role is a is a hybrid role for all intents and purposes, right? It I think organizations will have different thoughts on where they place a team like a team like mine. And it can be it can be in service, it can be in technology, it can be in a transformational unit of some description. But I think where it really started for us was with the expertise already in the team. So I had people that had worked in the in in the service center and people that were customer experience specialists but had really strong organizational knowledge as well because of their because of their tenure, uh, that had also had uh some great uh predisposition to be able to work with technology, interest in technology, uh, and an interest in automation, stuff like RPA stuff, and uh just had a real interest.

SPEAKER_00So it kind of we created a group of those people ultimately that were like journey managers, customer experience experts, uh customer CX analysts, and then started to kind of put real problems to them around the customer experience. So, how do we fix these journeys? What do we, you know, uh what can we build to solve some of these business problems? And then I think what evolved over the time was a real specialism for both being a journey architect or CX expert, but also being able, so for example, a lot of my team are going through their AI apprenticeship now with a company called Cambridge Sparks, so they're they're kind of AI qualifications where they get into prompt engineering, have to, you know, do a use a real life practical example within the business.

SPEAKER_00And obviously, we've got many, lots of them are centered around the use of polyai. And so, really, kind of it's about getting that balance right and and using what's what exists within the organization and then sprinkling in the expertise and and tech know-how in and around that to to round out the team. And that has happened, it happened relatively organically, but when we realized it was happening, we we wanted to then accelerate it and use it as a real USP for the team. And the results have been have been incredible, right? I think we've managed to build out and deliver more transformation than than most other parts of the business have in you know in the in a relatively short space of time.

SPEAKER_00Um, you know, that's been recognized within the business, and we've kind of been moved closer now to the teams that can help enable us, you know, in a speedier way. So we've moved into technology. I now report to the CTO, Tim, and you know, that's really to take any of the restraints off the team and how quickly they can deliver optimization back into the business using tools like Poly AI as kind of like our headline, uh headline game-changing tool. And that has then opened up the doors for these evolutionary conversations about our use of of Poly AI, from being a voice channel solution into okay, how can we use it within the marketing process and customer engagement lifecycle?

SPEAKER_00How can we use it within the product experience in our applications? How can we use it in terms of its integration with our financial systems and then use it to almost get into some conversations around MCP and they're like deeper integrations around how we might be able to kind of have poly in the box for some of our some of our systems? So we're we're getting the the conversations really quickly, you know, exploded into a a fest of the business saying, Oh, we we've thought this would be really interesting. We thought this would be really interesting. It's kind of like, okay, we we need to we need to kind of manage this a little bit, but it's really, really exciting, really exciting.

SPEAKER_01So look, I mean, I think it's like the most exciting era to work on anything that is like technology adjacent. And I think like what I've gotten a lot of personal pleasure from is, you know, through our early years as we worked with contact center teams, you know, there was always this like thought of like, hey, we are creating more interesting careers for people orchestrating it. But then it was also like, you know, what will happen to the headcount that was always like in the back of our minds. But the truth is, like, the work was never really getting done fully. And like the Jevin's paradox is like, we're just insatiable as humanity. We want more of everything, right?

SPEAKER_01So you just get to build like a bad thing with it. But as I looked at just like, you know, can like the next COO come from like a contact center, I think like, you know, the most AI-pilled kids of today will be like revising through the orgs fastest. And I think that is the areas where there is enough data for them to do incredible things that will produce the next like leaders. And I think finally now in the contact center, this is now where you know you have someone, you know, 22-year-old using Ren, just you know, sending two or three thousand messages back and forth.

SPEAKER_01And you know, when I look at cloud code usage, even our engineering teams, like in terms of AI pilling and learning, I think some of our customers, I think like when we were preparing for this, I checked her dashboard and uh Amy Corfund from your team is I think either number seven or number three in our entire customer base and just like voracious consumption of the tool and like you know, using it to make changes and iterate. It is just like so fun, and like the effort is so much lower, so then you can actually do all of that you ever wanted to, right?

SPEAKER_00You can experiment, right? Uh and and you can try things and you can learn through that that process. And I think I remember being a business analyst years ago, too long ago, decades, and um and learning how to write SaaS equal SPSS and and the most the the best way I I I got to learn all of that was by by Trying and doing and failing and learning and stuff. And I think this is this is where kind of the kind of more low-code, no-code solutions at the front end of your platform are really appealing to those people that have that have that kind of attitude and and uh ambition as well to try those things.

SPEAKER_00And I think it you're right, it offers opportunity. And and and I would say kind of my advice to CX leaders would be to kind of you don't need to tell your advisors that they need to become software engineers, right? You can start with like adjacent capabilities they already have, like conversation design, knowledge, QA, customer insight, root cause analysis, journey improvement, and then add the technology school skills around those strengths and allow them the freedom and the tooling to be able to try, experiment, learn, and support them in that way.

SPEAKER_00And and as we, you know, as we've learnt ourselves and and we've learned in our in our leadership team here at Simply Health that the results can be incredible if you can do that and if you can do that organically even better.

SPEAKER_01Can you can you tell us a bit about that? Just kind of like you know, how how did those evolve, how did you decide on the roadmap around channels and like expanding the capabilities?

SPEAKER_00Yeah, I think it's really interesting. So yeah, I think as I alluded to earlier, we kind of started out on the Boys Channel, right? And then we I think, and this is kudos to Polly AI, I think, because our partnership's been so strong, we've achieved some great results, and kind of partnership and deployment team has been brilliant in helping us identify those use cases. It was always inevitably going to raise the question of okay, well, how do we go faster? The more time we spent with the product, the more we learn about how it works.

SPEAKER_00And then, you know, subsequently our migration to uh to the Jupyter platform and and use of uh you know, use of uh some of the more enhanced tooling and no-co solutions kind of have fed into that desire of well, we could do this, right? And it's or we we could do this. So one of our uh ambitions uh in our kind of uh web uh our web and phone applications is to have a healthcare concierge. So this is a product design thinking, this is something that helps people answer real-world questions rather than just giving them a product carousel and kind of letting them choose and and have their own uh you know, leaving them to their own devices.

SPEAKER_00This is more about surfacing answers to people that have questions and in a at a point of need, right? And so when we kind of started to learn from product that that was the product design ambition, and they wanted to do some testing and build that out. My team immediately, well, we have some of these capabilities within a poly AI platform, right? How about we bring those to bear to that, you know, to that product design thinking and look at how we can build something? Poly AI is here, it's our platform, it's it's on estate, it's the capability is here.

SPEAKER_00We can build, we can experiment, we can A B test, we can auto-QA, we can do all of this the stuff that means quick deployment, all the things product teams love to hear, right? And so, and so immediately I had that first conversation with one of my peers, our head of health plan product, and he was incredibly interested, as you'd expect. He was kind of like, okay, let's talk about this some more, let's understand kind of how we can do that. And you know, do we want an SDK? Do we, you

unknownknow,

SPEAKER_00do we want to go headless? What do we want to do in terms of of of how we get this functionality over? And we're having that conversation right now with yourselves and with our product team. Um, and that's just one example, right? We've since got into sales and sales support, and because our remit as a team has expanded out of service now to being the entire organization. So, how do we put optimized, how do we optimize the organization? The product's moving so fast over here. How do we take the as is organization and match that pace? And that's really what my my team are there to do. And poly AI is one of the main tools in our toolkit to do that.

SPEAKER_00So, sales, we're now talking about how we can assist the sales process with Poly AI in various channels, but also with Next Best Action and expanding those capabilities out. So, chat and WhatsApp being a great tool for sales teams, you know, especially B2B, WhatsApp is an interesting one because it allows people to think they have an open conversation with an organization and gives a bit of a concierge feel, but is great for resource, right? A lot easier to resource, it's asynchronous. So yeah, you can see there, I've already touched on another plot things that we're trying to which we're starting to look at.

SPEAKER_00So I think people's natural curiosity because of the positivity around polyai in the business has led to lots of people starting to knock on the door and say, could we do this or could we do this? And then we had that conversation with our brilliant account team, and they say, We're not sure. Give us a few, give us a few days and we'll come back to you. And they could they do, and they say, Well, we could try this and we can and we're starting to co-create solutions together now, right? Which is a brilliant place to be in, right, for a partnership. It's a fantastic place to be in.

SPEAKER_01No, no, no. Look, I mean, I think we've been like tremendously successful, and you guys pushed us as well. But you know, I think that kind of like especially like that use of RET as a coding agent has allowed like the abundance of work on both sides of the kind of collaboration, where we're not, you know, the same way that the contact center was constrained, I think. And do you have enough people to just like do failure management and like pick up the phone call? You solve that, waiting times are zero, you get some time back. You get some time back for people to actually then do more of this kind of work, right? And then, like with a coding tool, you lower the cost of doing any of that work, and you don't need like a massive bill for the IT to do anything.

SPEAKER_01And then you know the iterations accelerate, and next thing you know, you're not just doing that like first call and failure management, but you're really kind of like wrapping yourself around the journey. And I don't think there's any rule or up, it all has to be AI or humans, but I think it's exactly that. Like, what do people actually want? Right. And if you can afford to experiment, then you find the better thing and then you crush your competitor because you have a better experience.

SPEAKER_00Correct. What's the customer effort in this? What is it that they are trying to do or seek an answer for? And and if you start to journey design not around how do we resource demand, but around what should the ideal customer journey be? And if you think about some of the transactional questions that people want to answer and they want an answer at their fingertips, well, think then about the channel that you want to have that interaction in and where they want to receive that information. And then think about also the journeys that you do want a human, you know, service, like bereavement, for example, or somewhere where you do want empathy and you want, you know, multiple activities happening on the back of an interaction that requires more skill, more depth of thought, one more contextual and emotive understanding, all that kind of stuff.

SPEAKER_00And design that way rather than the traditional CX, which is like, you know, the Pareto and give, you know, give me your volume and your failure demand, and we'll just work from the top downwards. Now it's become about architecture, you know, experience architecture and how AI and automation tooling complement that that architecture. I think that's how the gains changed for me. I think certainly over the and accelerated over the last few years, that's for sure.

SPEAKER_01Yeah, yeah. You meant you mentioned the hackathon that you should do. What were the use cases that ended up being built?

SPEAKER_00Yeah, so we chose a practice. This was part of like a team day across technology, right? And then we broke out into our rooms and myself and my team had a practi had a selection of practical problems we could solve, but we chose a practical problem within our practice support team, which is how can we identify. So this is in the den plan part of our business. So dentists, when I say practice, I mean dental practice, sorry. So how do we

SPEAKER_01very happy customer for about eight years now? So

SPEAKER_00happy to hear that. Happy to hear that. Well, you know my contact details if you've got any problem. Don't be bypassing poly A either when you call, right? Don't be hitting those speak to prompts, Nicola.

SPEAKER_01Never

SPEAKER_00cool. Yeah. So we picked uh we picked a problem in our practice support team around ID and V, which is identifying and verifying practices. And we know it works in our Simply Health agent, but we've not really explored it a huge amount over in the dental practice support teams. Uh so in in uh Simply Health, for example, it's achieving 85 to 90 percent success, which is brilliant. Uh and uh we identified that over in in the Demplan part of the business, it takes around 30 seconds per call to complete that. So, to keep it quite simple for the hackathon, we wanted to introduce an ID and V solution, so identify and verify solution for that practice support team to take potential routine work away from the advisor.

SPEAKER_00So then we used REN, we uh looked at the existing health plan journeys, we looked at how we would need to repoint some of the uh prompting to to Demplan and then and kind of the data sources there and validate some of that. We looked at the current FAQs that existed in Demplan, the dentistry part of our business, uh, and we looked at the current journey and any sort of supporting systems. So we had various parts of the team in the room doing those various different activities, and then so we built a a bit of an MVP by repurposing some of our health bottom prompting and re-engineering it to face into the other uh the Demplan part of the business.

SPEAKER_00And we then started to uh do a bit of live QAing on it and running some scenarios and trying calling up the phone lines and trying different stuff ourselves. Within the afternoon, we built out a feasible solution that has since sailed through all the various QAs and been tested and is going into deployment next week. So, an afternoon of the team's time, doing a bit of team bonding, I think has has built a practical solution. And I think that was a really interesting approach to solving that business problem. But it was great at seeing all the different parts of my team working together using using Ren and using Polyai to build out the solution.

SPEAKER_00I think the exciting thing for me isn't about the 30 seconds you know ID and V save, which is brilliant, right, from a business perspective, but it's the operating model behind it. It's the team spotting the opportunity, being increasingly capable through Ren of analysing that problem, prompting something, testing it, putting it into deployment, again using Ren to help us understand and complete those stages, and doing it in a encapsulated piece of time, right? A time-bound exercise.

SPEAKER_00I think that was pretty cool and a really interesting exercise. I think other organizations that are in partnership with Poly AI should should try the same thing. I think it's it's quite quite a cool, cool exercise to do.

SPEAKER_01No, it's incredible. And I think, you know, the one thing that like gets me going here is like it's just fun. And you know, when you kind of like get someone like an ability to have fun doing the work they do and probably already love, but then when you amplify it with this level of fun in it, you know you're onto something really good.

SPEAKER_00Yeah, yeah.

SPEAKER_01And like forget the business value, it's great that it pays for itself if it's at all successful and it by and large is and continues to be in like greater and greater, like you know, incidence rates, but it's more just like it's really fun, right?

SPEAKER_00Yeah,

SPEAKER_01and like it also like completely closes that loop between like the builder is the person who knows what a good experience is. So it's kind of like just that communication of backup. You build me this, oh, you don't get it, you didn't do this, you said this the wrong way. How is this not obvious to you? You know, that classical like business and IT teams pointing fingers and blaming each other. It's like, hey, go to your IT team to make sure that it works with like a huge workload that it's provisioned the right way, that it's safe, that it's all the things that those guys are actually like they have fun doing that, right?

SPEAKER_01Because that's what wide are in like IT, right? Whereas like the people who are like in CX are there because they want to create great experiences. So yeah,

SPEAKER_00you might as well cut out

SPEAKER_01a bottleneck, right?

SPEAKER_00Absolutely correct. And and I think that's been part of our move into tech. And uh, and what was great in the room during the hackathon, right, was that we had some of the team haven't used poly as much as others, but it was actually contagious that the I want to give this a try, right? When can I come in?

SPEAKER_01Yeah,

SPEAKER_00and so and I think as you've alluded to there, right? The the the the safety elements and the release protocols and the stuff that our tech teams own, and rightly so. Having moved into technology now, we're closer to that, right? And we can we're kind of joining up the party a bit here between the the people that can understand poor business operations and poor process and inefficient process and failure demand, but also customer experience and and championing the customer experience, closer to the people that can facilitate the solutions, it creates a pretty powerful business unit, I would say.

SPEAKER_00So, yeah, a few a few things there that we that we took out of that exercise, that's for sure.

SPEAKER_01Yeah, awesome. Well, look, it was really, really good to have you on here. Really excited to hear about all the moments, and thank you for sharing them with with the audience as well. I'm really excited about just like seeing what you guys will build over the next period. I feel like you

unknownknow,

SPEAKER_01time has now collapsed, and we'll probably build more in the next six months than we did in the last three years.

SPEAKER_00We've got some exciting things coming, that's for sure.

SPEAKER_01Yeah, yeah,

SPEAKER_00thank you. Thank

SPEAKER_01you. Thank you again so much our meet. It was great to have you for the viewers. Please like, share, subscribe, and we'll see you in the next one.

SPEAKER_00Brilliant. Thank you.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. What happens if your AI agents get 1% better every day?

    Sep 10, 202639 minAudio

    An agent that gets better every day compounds into a very different agent within months. That's the idea behind Wren, PolyAI's new release: an agent whose job is to make your other agents better. Some of what it finds...

    Credit the people on this episode

  2. Who's coordinating your army of AI agents?

    Sep 3, 202644 minAudio

    Companies are pouring money into AI, but not everyone is getting the returns. Ian Jacobs, VP and Lead Analyst at Opus Research, thinks he knows why. In this episode, your host Nikola Mrkšić sits down with Ian to work...

    Credit the people on this episode

  3. Why should CX leaders care about MCP?

    Aug 20, 202636 minAudio

    When PolyAI built its Builder MCP, one client helped shape it: Vixxo Facility Solutions, one of the largest facilities providers in the US. In this episode, guest host Damian Sasso talks with Vixxo's CTO Derek Neighbors...

    Credit the people on this episode

  4. Can AI really hear a call the way a person does?

    Aug 6, 202622 minAudio

    PolyAI just launched Dialog-RSN-1, its first audio-native model, and your host Nikola Mrkšić sat down with its builder, Matt Henderson, to unpack why it’s a game-changer for building voice agents. Most voice AI either...

    Credit the people on this episode

  5. Is word error rate just a vanity metric?

    Jul 21, 202629 minAudio

    Voice AI vendors love to quote word error rate, usually somewhere around 2 to 3%, as proof their system understands customers. Oliver Shoulson, Agent Design & Engineering Lead at PolyAI, thinks it's one of the most...

    Credit the people on this episode