How Triple Point Replaced an HR SaaS Tool with AI in 6 Weeks

Charlie Cowan: …and talk about something that is top of mind in the world of AI. Either stuff that we're doing internally at Kowalah, stuff that we're doing with clients. And I'm absolutely thrilled today to welcome Shelley Davison from Triple Point. We're going to be talking all about all things people and HR, and I'll ask Shelley to introduce herself in a moment. But for those of you that it's your first Wednesday webinar, we're in a Zoom event, and that means that, depending on your device, either at the top of your screen or the bottom of your screen, you're going to have a chat bar, or a Q&A, and I'm going to endeavour to keep my eye on these, so if you've got a question as we go through, then feel free to dive in. We're gonna follow our normal process.

Charlie Cowan: So we'll spend, I think today, probably the bulk of it, 45, 50 minutes, talking about our main topic of the day. We then talk a little bit about AI in the news, and I only really got one topic that I might touch on, depending on how our conversation goes. And then, as I said, we've got Q&A, but don't feel that you need to wait until the end of the session. If you've got something about anything that Shelley or I are talking about, then feel free to ask your question, and we will come to it. So with that, I'll just introduce a little bit of Triple Point, then I'll ask Shelley to introduce herself. Triple Point is a London-based financial services company that we've had the pleasure of working with for just over a year now.

Charlie Cowan: Around, I want to say, around 250 or so people in the team, helping individual and corporate investors to invest, and certainly for those individual investors, to invest wisely, for tax implications. And very much over the last year, the Triple Point leaders, the managing partners, have been injecting AI across all parts of the business. And we're thrilled to share one of those stories here. With that, Shelley, it would be great for you to introduce yourself. I know you're relatively new into the Triple Point team, but maybe a little bit about your background and your role at Triple Point.

Shelley Davison: Sure, hi, I'm Shelley. I'm the People Operations Director at Triple Point. I've been with Triple Point for about 5 or 6 months now, and I'm responsible for all of the functions within the people team, so talent acquisition, learning and development, business partnering and operations. And obviously I had the pleasure of working with Kowalah to build this system. Prior to Triple Point, my background has been in various industries, construction and education, but all within the people space, always had involvement with systems integration and implementation, so, yeah, this project is very close to my heart.

Charlie Cowan: Awesome, brilliant. And maybe just to sort of set the scene of the conversation that we're gonna have, yes, you know, this is in the realms of AI, but our goal here is not to either teach people how to build something in AI, it's not to dive into any of the technical side or the architecture. Shelley and the other members of the people team are coming to this as a people team. And, as we're having our conversation, we're speaking to you, dear listener, who we believe is someone that's also in a people team, thinking about, oh, well, I'm not a coder, I'm not an engineer, how would I even go on a journey like this? And so that's very much the frame of the conversation that we're gonna have. So maybe let's go back to the start, and I remember it was January the 16th when I got an email from one of the managing partners going, hey, Charlie, when are you next back in the office? We've got an idea. And I think you had either just joined, or maybe you hadn't even joined yet, but can you cast your mind back to that, at the very start of the process?

Shelley Davison: Yeah, so I hadn't joined at that point, so I joined on the 19th, so a little bit later. And actually, obviously, the first couple of weeks was always a bit of a whirlwind, so it was actually more like the beginning of February, where I was brought into the loop that we had this idea, or there was this instruction that we were removing one of the systems that we had previously used, and we had a very short period of time to replace that system using AI. Do you want me to go into detail about what those systems were, Charlie, at this point?

Charlie Cowan: Yeah, I think it's good because anyone that's worked in a people team, an HR team, has probably had this challenge of getting feedback, so maybe you can talk a little bit about that. Obviously, this system wasn't there when you were there, but you've worked in people teams before, and trying to get this feedback, so what's been your experience and your challenges?

Shelley Davison: Sure, so at Triple Point, the previous system that was being replaced was to gather insights from around the business, from peers around who people think were the high performers, or an opportunity for people to call out those that have demonstrated our values, living our values. So it was really almost like a recognition system, so it just brought people to the forefront so that the leadership team could see who was being nominated. And then separate to that, there was another method of gathering feedback which fed into the performance appraisal. So that was your typical kind of 360, albeit there was no system in place for that. So what was happening before was managers were instructed to send various emails across the business to people that they chose, to get feedback on certain employees.

Shelley Davison: So, obviously, they had to proactively manage that, they had to send lots of emails, and then manage lots of emails coming back in, and sort all the feedback, and try and make head or tail of it, so that they could use that feedback on that individual for their performance review. So that was the system or the process that was in place. Now, obviously, the system to gather those nominations, the decision was taken to stop with that, so we had to find a solution, and we wanted to build in a solution for the multiple emails and the gathering of feedback, as it was very clunky, very time-consuming. So, obviously, the project scope was, look at both, build a system within a short space of time, which I must admit, I was skeptical whether it was possible, whether we could do it, and actually, my mind was going to, we might need an interim solution.

Shelley Davison: We're probably not going to have a system ready for March, but we'll have an interim solution, and we'll have a plan for a system. And that's, you know, that's where the project began. So, obviously, Kowalah very kindly came in to help us with that. And the rest is history!

Charlie Cowan: Yeah, I mean, I have been an employee in a number of companies over my career, and whether it was feedback and some nominations on other people, whether it was personal development plans, whether it was setting your goals, I've just been faced with a lot of these forms, like, just fill this thing in. And not just in the people team, this could be salespeople filling in CRM systems, it could be, I don't know, finance teams filling in orders, or whatever it might be. As humans, we're just not very good at filling in blank forms. And I think that's something that feeds into everything that's gonna come later on in our chat, was how do we make it easier for, people generally want to give feedback, but they just don't know how, and so how do we get away from this world of having blank forms, blank fields that people need to fill in?

Shelley Davison: I think as well, Charlie, that the challenge that I've faced in previous companies with these kind of processes is it's very typical to have lots of disparate systems, so everywhere I've worked previously, we've had a separate system for people to recognize their peers, have had a separate system for 360-degree feedback, a separate system for performance reviews. So that can create quite a bit of frustration for the managers, because, they're busy anyway with their day jobs, and then on top of that, annual review time comes around, and then they've got all these systems, and all these forms, and all these things they need to do, so that can be quite clunky.

Charlie Cowan: Yeah, yeah. So, to set the scene then, so I had the first email on the 16th of January, you joined 3 days later, and this initial email was, you know, we've got this feedback system, it's costing us X thousand a year, we don't think we need to be doing that for people to fill in a form badly once or twice a year. We think that with some of the stuff that you guys are doing with AI, we can come up with something better. So that was kind of the original sort of brief. So what we then did, which was, let's say, a couple of weeks later, so getting into the start of February, to get the conversation going, we wanted to not focus it on trying to build something. At the time, ChatGPT was the main tool that was rolled out at Triple Point.

Charlie Cowan: And so what we did was to set up a shared project in ChatGPT. If you've not done one of these before, you create a project, you give it some information, and then multiple people can use this project as a source. And, so what we did was feed this project with all of the background from my initial conversations with the managing partner and some of the team about the current situation. And so you can see the project instructions here. We have got a project goal to create a comprehensive product requirements document for an AI-powered employee feedback system, replacing the current one. So the key of this exercise was, right, we want to get the five or six of us around the table, in the Triple Point office, and this is a non-technical session. It's just about understanding what the solution should do.

Charlie Cowan: And so you can see here in this conversation that what we did, we had the big plasma screen up in the meeting, we had the six of us around the table, and then we were chatting with ChatGPT as if ChatGPT was a member of the team. So you can see here, it says, hey ChatGPT, because I was using voice, because I always use voice. Hey, ChatGPT, okay, we're now in the meeting room, we've got the team here, we want to start thinking through what we're going to design for this feedback tool. Can you help us to think through this? Firstly, can you take a look at all of the documents that have been added in? Because the team had been adding in some screenshots of the existing system. Can you get yourself familiar, and then we can work out what we should discuss first.

Charlie Cowan: So it really was, like, having a meeting, and ChatGPT was one part of it. Shelley, I don't know what your sort of experience was as we started having a bit of this conversation, or whether this was a very surreal thing, having a conversation with ChatGPT.

Shelley Davison: Absolutely, yes. So, obviously, I've used ChatGPT before then, never used the voice, so, yeah, that was all very new to me. But it was, I mean, it was a great process in that we could kind of just sit and talk to it and ask it what we wanted it to do. The thing that really stood out to me is every time we asked you, Charlie, can it do this, or could it do this? And we'd get excited about various things, the answer was always, it can do anything, which was really exciting and really empowering. But yes, it really did take me by surprise, because I'd never experienced AI in this way.

Charlie Cowan: Just having it, yeah, yeah. And one of the analogies that I use is of crazy golf. So for anyone watching that's played crazy golf, you'll know, you're standing here with your golf club, and then you want to get it in the hole, but you can't, because there's a, I don't know, a windmill or a dog in the way. So you have to aim off to the side, and then bounce off, and that's how you're gonna get into the hole. And this is very much the same thing with designing and defining the things that you want to build with AI. The temptation is to ask, build me a new HR feedback system. That would be aiming your ball straight at the hole, but instead, what we're asking here is, we want to build an HR feedback system, can you help me to think through some of the requirements that we might want?

Charlie Cowan: And what that might look like. And so that's aiming sort of indirectly, and we find you get a really rich conversation from that. The other thing that I'd add, so you can picture the room now, so you've got the plasma screen up, you've got the six people around, I'm now standing up, next to everyone, next to the whiteboard, and I'm drawing on it. So as the conversation is developing, that the team are speaking to ChatGPT and answering questions, as they're saying, oh, it could do this, it could do that, and this is what the current system did, then I'm drawing that. And so this is just classic, good old, I don't know, consulting, business and analysis, sort of teasing out, asking questions, you know, when someone says something that you don't know.

Charlie Cowan: And what I certainly found at leading the session is that everyone starts gravitating towards looking at this whiteboard, going, no, that's not quite right, and I was rubbing a few things out, and… So, how did you find the visual nature of bringing what your ideas were to life on the board?

Shelley Davison: Yeah, it was helpful. I think what you can see on the screen is obviously the two different systems. We were trying to create a solution that fixed two different processes. So it was helpful trying to separate that, so I think having the ideas on the board was really good. I think what was really helpful, actually, I think, as you were saying at the beginning, you know, you think you can just hit the golf ball straight into the hole, and so it started off, well, we want a feedback system, your typical feedback system, we want it to do this, and actually, when you then started talking about the different things we wanted it to do, or the different realities that we had as a business, it really generated much more discussion, so it became, it went from this kind of simple thing to, you know, much more detailed and much more tailored to our process. So obviously, as the workshop or the session continued, the energy just rose in the team as we realized we could do more and more.

Charlie Cowan: Yeah, yeah. And I remember, I mean, I'm looking at the 360 on the left here, and then, we made the decision, well, actually, no, we don't want to do a full 360. And actually, on the right-hand side here, business impact, this is over and above the performance review. It's not the performance review process. And so actually visualizing some of this stuff, things that were, I don't know, requirements at the start of the conversation quickly became not requirements. I was like, oh, actually, no, because now that I can see it, maybe this isn't exactly what we want, which is where the crazy golf comes in, because if you just asked for what you wanted up front, then you end up not getting what you want. Nightmare.

Charlie Cowan: And so then, this was, we love doing this in these workshops, so one is, like, drawing something on the board and taking a picture of it. And increasingly, we're using Granola as a call transcription tool to listen in the background. We weren't actually using it in this one, but if you can imagine, Granola takes the full call transcript. And then, what we said here, can you take a look at this picture? Let me know if there's anything missing, or you can't read or understand, and then we can start to frame out our product requirements document. So we were in the room, and we did this workshop on the whiteboard. We then took the picture of that, gave it to ChatGPT, and said, can you see what's on here? So it says here, yes, I can see it clearly, it's actually very helpful. I'll show you this in three parts.

Charlie Cowan: What I can read, what I'm interpreting, any ambiguities, and so on. So what I can read, left-hand side is the core of the 360, and so on. And what was wonderful, and this was great in the room, because everyone was like, it can see, it can see! And, just such a great way of getting quite complex discussion back into the system.

Shelley Davison: Yeah, and do you know what? That's always the frustration, isn't it? When you do these sessions, and they're great discussion, you enjoy the session, and then someone has to write up the minutes and interpret it and make it happen, whereas it was just happening real time, which was.

Charlie Cowan: Yeah, take a picture, there you go. Tell me if you don't understand anything. So, that was the first session, and maybe we spent, maybe that was a two-hour meeting that we had, and that would have been on the Wednesday. We then met up the following Tuesday, and a version of what we've got in front of us was what we walked through. And, we built this using Claude Code, so ChatGPT was using the projects to get the product requirements document. But what that gave us as an output was, I want to say, like, a 5- or 6-page document that had all of the different sort of features and functionalities of this tool, who was going to use it, how they were going to use it, what the cadence was going to be of getting the feedback. And what we did was feed that into a new Claude Code project.

Charlie Cowan: We pointed Claude at the Triple Point website, which had just been through a brand refresh, so it now had all these lovely new sort of colors and fonts. And Claude was able to bring together this first version of this HR feedback tool. And then on the following Tuesday, we're able to get a first sort of click-through with the team to go, right, is this along the right lines? And so maybe, Shelley, just a bit of your sort of experience of, yeah, within a week, being able to go, right, okay, I can start to see our ideas come to life.

Shelley Davison: Yeah, I mean, the session was fun, and it really helped us to kind of rationalize what we wanted for our process and what would work for our business. But it was still hypothetical. It was still kind of great, okay, but how is this going to work, and how are we going to do this in a number of weeks? We still had that skepticism. And then when we saw the system, it was very exciting, because suddenly there was a thing, there was a tangible output to the meeting, which then we could refine, and we could test, and we could see. But yeah, the team and I were so excited when we saw this. It was amazing how quickly that came up.

Charlie Cowan: Yeah, and I think there were two things that I took away from that first, well, the second meeting, but when the first time of having a look at this was, because Claude had had a look at your guys' website, it felt like a Triple Point application, not like a standard bit of SaaS that, alright, well, this has been built for everyone, it's just a new product that we're using. There felt like there was a bit of ownership, like the font was your font, the colors were your colors. So that was one thing. And then the second, and as you alluded to this earlier, was then the, can it do this, question started coming out in full force. Right, now I've seen what's possible in 5 days. Can it do this? Can it do this? Can it do this? To which I, yeah, jokingly said, it can do anything.

Charlie Cowan: The question actually is, should it do? So, yes, it can have a button that does this, yes, it can have a button that does that, but should it, and what's the user experience? And that was really nice, to then tie this back to an employee who's being asked to give feedback, how do we give them a great experience, a manager that's got to review all of this, and then you as a people team, and the senior execs.

Shelley Davison: Agreed, and do you know what, actually, when we had the system up and running, it was then very easy, because we could play with the system, we could then say, oh, actually, no, we don't want to do that, and we thought of things that, you know, issues that might have popped up, or other things that it could do. So I think that initial getting it up and running, get to that stage as quick as you can, because when you're there and you can see it, it all becomes very clear what it can do. I think, like you said, the danger, I think, for our team, particularly me, was, we wanted to do this, and we wanted to do this, and oh my god, it can connect to this and this. And I think that there was multiple times where the conversation was, right, we have four weeks, we have to do a phase one, and we can think about some phase two. So I think that's always, if you've got a tight deadline, just make sure that you limit yourself to what's realistic in the first phase, and there's always opportunity afterwards to develop it, but just don't try and do too much.

Charlie Cowan: Yeah, yeah.

Shelley Davison: This is my advice.

Charlie Cowan: And on deadlines, I was, when I was pulling these slides together, I was, oh, yes, yes. So when I go back into the original meeting timeline, 6-week development window to support a 24th of March performance review cycle, and then somehow, which is all fine when we look back on it, but somehow, after this meeting, it was then the 2nd of March, so things moved forward a little bit, and maybe that was just sort of the excitement that we can build something really quickly here.

Shelley Davison: Well, the performance review cycle starts on the 24th of March.

Charlie Cowan: And then this was to lead into that.

Shelley Davison: A couple of weeks before, so we've had the data to feed into that system.

Charlie Cowan: Nice, okay.

Shelley Davison: Slightly misleading for you, Charlie.

Charlie Cowan: Exactly, exactly. No, perfect. And so, we're not going to do a demo of this, but I thought it'd be useful to just talk through some of the elements, and what we wanted to inject into this is, it's not just using AI to build something. I say that's great, because you can use it for the product requirements, you can use it for the build and testing and so on, but what we really wanted to do was to inject AI into the heart of the experience for the users. Because no one likes filling in a form, and if all we've done is create a new system where people have to fill in a form, then that is not great. So, there were really two elements that an employee would be asked to do, and you can see them on the screen here.

Charlie Cowan: So, one was to provide feedback on other people where they've been nominated to provide that feedback. So, as an individual, you with your manager might select 5 or 6 people to give feedback on you. And so, as an employee, you'd come to the feedback tab, and you'd say, right, these 8 people, you need to give feedback. The second was insights, which is, if feedback is requested from you, insights is more going out. So, who do you want to nominate who exhibits this value, or whatever? So maybe we just talk about the feedback section. We've all been on the receiving end of these forms where you just have to fill something in and you're not quite sure. We built this on the Anthropic API, and we provided a skill behind the feedback, which basically was loaded up with all of the types of questions that we wanted to tease out from someone.

Charlie Cowan: And so when someone had to give feedback, let's say they were giving feedback on Shelley, it would say, great, a kind of chat interface, you're going to give feedback on Shelley, so tell me a little bit about how you work with Shelley, tell me a little bit about how she's really helped you out this year. And so it was more of a conversation that was led by Claude in the background. So what was your sort of experience? Because, obviously, you built the thing, but you were providing feedback on other people.

Shelley Davison: Yeah, I mean, it's probably not a fair test, because by this point, I had looked at this system and developed the system every day for the last four to six weeks. But no, it was very quick and easy to use. It was obviously, I don't know how much detail you want to go into about the AI tool, but it was very intuitive, it asked the right questions. And actually, when you're filling out a form, there's a lot of care that you have to take over making sure that it sounds professional, or you've made it sound perfect, but actually, you didn't really need to do that, you just had to talk to AI, in the way that you would normally talk, so it felt very much like just having a conversation rather than this admin exercise that you had to do, and the system was very easy to use and so intuitive. As I say, I'm biased, but it felt a very easy process.

Charlie Cowan: Yeah. I remember we were doing some of the fine-tuning was about how long do we want one of these conversations to be? So, at the very short end, you could imagine it just saying, do you like working with Shelley? Give me one example of how she was great. Give me one development experience, very sort of short and to the point, and that's, well, that's like filling in the form, that you're no further ahead. Then the other extent was actually this turned into more of a sort of a coaching session that felt like there was no end to it, and someone's like, oh, I could be here for 30 minutes having a chat, and I've got 10 of these people. So there was a bit of fine-tuning of the prompt and the skill to get to a point where it felt like a conversation, but I think we were aiming for maybe 5 to 10 minutes maximum for each person. You could have a 5-minute conversation and you've given your feedback.

Shelley Davison: Yeah, and actually, that was the most challenging bit, wasn't it? In that you want certain data to come out of it. You want consistent output so that it can be useful to the managers, and we can have consistency across the board. So you need an element of structure, and certain things being pulled out of the manager or the employee that is filling it in. But at the same time, like you say, you don't want too much structure, that you might as well just have a form, so it's just trying to strike that balance between being conversational, but also getting good quality data and consistent data that came out of it. But actually, what was great was we didn't have to find that solution in the meeting. We didn't have to say, well, how are we going to do that?

Shelley Davison: All we had to do was vocalize what wasn't right with it, how we wanted it to be tweaked, we want it shorter, less probing questions, a bit more structure. And that was kind of the level of detail that you have to give Claude, and then Claude would come up with a solution. It was, like, having a really intelligent colleague in the room to help guide you through it. So you, if you get stuck or you're not sure, that was so helpful that we could just ask Claude, and Claude would tell us.

Charlie Cowan: Yeah, yeah, yeah. I had a lovely chat with one of the senior leaders at Triple Point a few weeks or a month or so afterwards, and he was saying, look, as a senior leader, because we did this kind of 360, or we call it 180, but, you could be on the receiving end of having to provide feedback to 25 or 30 people, which, on the one hand, is lovely. It's lovely that people want your perspective and that you're touching on their careers, but the downside of that is, right, well, if I'm going to spend an hour on each one, my goodness, I'm going to spend a whole week providing feedback.

Charlie Cowan: Now, we had built a bit of a voice element into the platform, so you could, through your browser, you could talk to it, but what he did was actually, sort of slightly outside of this platform, once he knew what the 3 or 4 questions were to provide feedback, he was using ChatGPT Voice, so with ChatGPT Voice, for those that don't know, you can talk and it talks back to you, you have a whole conversation. And he was in his car, driving to work across a few days, and was like, hey, ChatGPT, I want to provide my feedback around these 10 people. And so he was just having a conversation, ChatGPT's building out the transcript, and then he came back to the platform and was able to just drop it all in.

Charlie Cowan: And that was a really nice way of going around it, and something that we can build into the next iteration, so that people can really, just on their phones, provide this feedback directly into the platform. Next one I wanted to show is, yes, there's using AI to capture the feedback and the insights from people. But then, you're thinking again about the different sets of stakeholders and users of the platform. One of those is either the managers or the senior executives. So, across a company, you've got 300 bits of feedback that, well, 300 people or so, and so there's thousands of individual bits of feedback. And it's like, well, how am I supposed to make sense of this?

Charlie Cowan: Yes, maybe some poor person in the people team has got to go and build slides and spreadsheets and analyze all of this stuff, but hang on, we've got Anthropic baked into this thing, and all the data is there in a structured database. So what we built on the right-hand side for the executives was a separate chat widget for them, which has basically summarized the progress of the campaign, and so that was useful in terms of nudging managers that hadn't necessarily got all of their teams to nominate and get feedback, pulling out who are the best people for this value, who needs help, and so they're really able to dive in and use this data. One of the niggles that I've had with, I don't know, employee pulse surveys and things like that, is sometimes you think, well, hang on, I'm spending my time giving all this feedback, but who's using it?

Charlie Cowan: What are they doing with it? And this really helped the execs to bring the data to life.

Shelley Davison: Yeah, and the managers as well, on the feedback side, that was a really time-consuming process for them before, when they were having to receive emails, save the emails, copy and paste, review the information, come up with the summaries themselves, whereas it took all of that legwork away and just provided all the key themes, even once it's given the key themes, if the manager needed help on how to deliver certain messages, you could even ask it to do that, and it would help coach you through those discussions as well. So, it's a huge time saver.

Charlie Cowan: Oh, brilliant. Yeah, we had, we thought about different stakeholders in here, so there was the employees themselves, the managers, the senior executives, IT, and we'll come on to IT and sort of governance and security in a minute, but then also the people team. So they've all got different dashboards looking in at the same sort of system, but from their own lens. And so, talking about the people team, one of those lenses was this admin portal, which enables the people team to manage campaigns off into the future. And so, they could update and manage employees, or could plug this into the HR system directly to manage that, create and launch campaigns. But what we've got on the screen here are the Triple Point values. And behind each of these values are a certain number of high-performance indicators, that exhibit those values. And what we wanted to do was to give Shelley and the team the ability to own and manage and update these, that as values evolve, that they've got a sort of clicks interface to be able to update this. I don't know how you found that in terms of the ability to edit the campaign and manage it as a people team.

Shelley Davison: Yeah, I mean, it was so user-friendly. We could just go in and make changes. We did extend the deadline for a few days, so we're able to go in and do that. You could manage email notifications, how frequently, who gets them, what the wording was. So, we had total control over the admin side of things, and it wasn't, some systems you have to understand how to work the backend, and they're not the most user-friendly, whereas this was just really simple.

Charlie Cowan: Yeah. And, so we touched on IT, or digital and tech, as they're called in Triple Point. And we worked alongside one of your colleagues, Ben, who did a fantastic job of taking the sort of the initial idea and application, and bring it inside your four walls, so that it's governed and compliant. We're obviously dealing with very sort of private and secure personal data here. Anything from your experience, not coming from the digital and tech world, about any sort of advice or recommendations you'd have to someone else about embracing and working with their IT and tech teams?

Shelley Davison: Yeah, I mean, my first piece of advice, and, you know, if Ben is listening to this, he'll say, follow your own advice. Involve tech early, because obviously, in our situation, we did have a very uniquely tight deadline, so the usual planning and thinking and getting people bought in kind of went out the window a little bit. It wasn't your typical project. So what we didn't do was involve tech early enough. So, we got very carried away looking at the functionality and what it could do, and how amazing AI was, and then the reality hit of, oh, wow, we're actually due to launch this in a few weeks, and have we thought about data security? Where's this going to be hosted? How are we going to integrate it with our systems?

Shelley Davison: And so, we weren't very popular with Ben in the tech team, but we were very lucky to have him because he was amazing, equally excited about the project, so he was able to, well, he kind of dedicated the time, and we just made it happen. But I definitely wouldn't do that again. I wouldn't put that pressure on the digital team. Again, I would obviously involve them from day one. And not just because, obviously, they need time to plan, and they need to find the resource, but also because they were able to bring so much. Ben has brought so much into the project in terms of, have you thought about this? And there's things that we would never have even considered as people professionals. We didn't have that skill set. And so it was really helpful to have him.

Shelley Davison: We were very lucky he was able to coordinate it at such short notice, but definitely involve tech as early as possible.

Charlie Cowan: Yeah. I mean, we're definitely seeing this across companies, across different functions and teams. It's so easy to build stuff now, whether that is in Claude Code, or Replit, or Lovable, or V0, or Bolt, all these tools exist. And for a kind of citizen developer, someone that's a bit enthusiastic, that is in finance, or people teams, or operations, suddenly they've vibe-coded this application that plugs into their CRM data or their accounts, and I've had so many stories of people that didn't work closely enough with their IT team, their governance team, and suddenly, secure customer information or employee information is out there on the public internet. And so, it's great, and we want to encourage people to bring their ideas to life, like we've done like this, but to do it in a safe and governed way, so that no one gets caught out.

Charlie Cowan: And so I think that these were three things that sort of jumped out to me before we then talk about what happens next. So we've talked a little bit about bringing digital and tech in early. One is the governance idea, and that's important, we need to make sure that we're doing this in a safe and governed way. But increasingly, we're seeing this across our larger customers, people are building the same thing that someone else has already built. So, maybe it's not in the people team, but it's, salespeople building a forecasting app, or marketers building an event landing page. Someone's already built that, and so how do we have a consistent library where we can see this, and the digital and tech or the IT team can be a nice central place for that.

Charlie Cowan: The second thing, we talked about this right at the start with, having the whiteboard, having the creative session, is knowing what the process is going to be. Whatever you're building in, whether it's Claude Code or Codex or anything, that's actually the last place you want to go to. You want to work out exactly what it is first, because, like we talked about on the whiteboard, if you started building straight off, you'd have built the wrong thing. And then the third, which we talked about a little bit there, was, you don't know what you want until Charlie has shown it to you, and then you're like, now I want this, I want this, I want this.

Charlie Cowan: And so actually, the old world of doing something, which was, let's write a really long sort of specification and deliverables, well, actually, let's get something mocked up as quickly as possible, and then that will inform our ideas. Any other sort of lessons learned, or sort of best practice that you'd pass on to someone that was exploring this kind of thing from a people team, Shelley?

Shelley Davison: So, I think you've hit the nail on the head there. I think really understanding what you want from it is key first. As I said before, digital and tech, and seeing that first version is really helpful. So yeah, I think that my advice would be just start.

Charlie Cowan: Just kidding.

Shelley Davison: Going, just explore, just learn, with the support of Kowalah, you can really try and understand how it all works and how to do it, and I think as soon as you start, as soon as you see what's possible, that will, it was just self-perpetuating, then it will just be so exciting, and your possibilities are endless, so just…

Charlie Cowan: Yeah. Well, and that feeds into our last slide, so when Shelley and I were preparing for this webinar, I said, well, I hope you don't mind, but we've gone on from where you were. So, maybe you want to talk a little bit about what's on the screen now, and how you've got to this point, and what you see as the future.

Shelley Davison: Absolutely. So, obviously, working with you, Charlie, was really helpful, because not only did you build the system for us, but you were building it in the room. So, obviously, guiding us as you were doing it, telling us what you were doing, showing us how it was all working. So, we thought we'd be brave and build upon that. So, as a people team, we, and I'm sure many people in the people teams are doing, reviewing our massive tech stack, and looking at where we can not have so many systems, or how they can connect better. So that's something we've been looking at. And then, obviously, this project just really opened our eyes as to other solutions out there, rather than just off-the-shelf, multiple systems that can do what you want. Where we, when we were looking at our strategy and our aims for the year.

Shelley Davison: We kind of split them down into projects, and what we found was most of the projects we were working on involved some kind of automation of a process, or a system that we needed, or development of an existing system, but we were limited because our current systems don't have the functionality that we need that's bespoke to the business. So, we thought, why not build it ourselves, seeing as pretty much anything is possible. So what we've done is, what we didn't want to do, is pop up loads of different systems, and then exacerbate the problem of lots of systems, and hosting all these systems here, there, and everywhere. So we took the platform, the employee feedback platform that we already had, and we thought we would reconfigure that, so that it is just a people interface.

Shelley Davison: That employees could go to as a one-stop shop, and then the feedback tool would be sat integrated into that system. So you'll see on the screenshot, one of the buttons is to link to the feedback tool. But what this enabled us to do, or is going to enable us to do, is every time we want to build that system functionality, or automate a process, or share information, we can now put it onto this interface. So the next project that we will, we are working on, is we are due to be launching this within our business over the next few weeks, and we've automated our approval to hire process. But not just automated the process.

Shelley Davison: With a form, we have taken a similar approach to the feedback tool, in that we have built in an AI assessment, so the AI itself can help the business, sorry, help the manager build a business case. Going one step further, the AI is actually helping the manager discuss whether AI could do the role, or elements of the role as well, so we're really trying to embrace that side of things across the business, and so that's what we're working on there. And then our future plans, they're getting out of hand, I'm not gonna lie to you, because we're loving AI, but we're looking at talent modules, and then how that can feed into the feedback system. So, so yeah, so that's what we've done. It's not very pretty at the moment, because it's still in development, but we're all having great fun, seeing what it can do.

Charlie Cowan: That's so good, so good. Thanks for walking through that. Like, when we were going through, in February, March, and we were sat there with Claude Code up on the screen, and I remember us saying, you're probably thinking, well, why do we need to be seeing Claude Code in action? And it was because I want you to believe, to see, this is possible, we're just talking to Claude, and our ideas are coming to life. And we want to teach you to fish, not just to, right, here's a new HR feedback platform, because otherwise you're no further ahead than you were. And to see this is just absolutely brilliant, because it's like, you've got that belief, and, right, I'm going to go and build something with Ben and digital and tech, but we can bring any of these ideas to life now.

Shelley Davison: Absolutely, and you'll be pleased to know that Ben is very heavily involved from day one.

Charlie Cowan: Yeah. Fantastic. Well, Shelley, thanks so much. This has been super helpful, and I know that anyone that's watching live or on YouTube that is in a people team has been thinking, oh, well, I can't, I'm not a developer, I'm not technical, but I've got an idea. This gives them a little bit of that belief to, well, I'm just gonna see how far I can get, so thank you. Thank you very much.

Shelley Davison: Thank you for having me.

Charlie Cowan: And so with that, we are at, well, we've got a quarter of an hour, but I'm only going to say one thing on AI in the news before we wrap up, and that was just to highlight that this time last week, we were talking about Anthropic's launch, and then subsequent removal of Fable 5, this super powerful model, and it was just to say that there is no news on that. It is still not re-released. And we're still waiting to hear from that, so lots of discussions going on between Anthropic and the American administration. So we'll have to see when we get to this point next week, what that looks like. So to wrap up, at this time next week, we're going to be talking finance again, but a little bit more focused on private equity firms.

Charlie Cowan: We're doing quite a bit of work with PE firms, either for themselves, for the investment teams at PE firms, but also into their portfolio companies. The business model of a private equity firm is to inject some capital, to increase the valuation of that business, and then to realize that return at some point in the future. And AI is just an absolute godsend to that business model, because it's all about how do we grow a business without growing the cost base, and a lot of that is around, so, look forward to giving you some examples, things that we're seeing, and some tips and tricks, either if you are a PE firm, or if you're a portfolio company of a PE firm. With that, thanks very much. Thank you again, Shelley. Absolutely brilliant. Great to chat with you, great to work with you, and we'll catch up with everyone again this time next week.

Shelley Davison: Thanks, Charlie.