Verena: Welcome back, Simon. So how was your chat with Ben? Tell us about one key learning, please. If you enjoyed this episode, subscribe for more conversations like this. We also share practical Start, Launch, Grow insights in our newsletter. Thanks for listening and see you next time. Welcome back to the podcast, everyone. Today we are doing something slightly different. So usually Simon and I pick a topic, but this time we are answering two of your questions.
Simon: Ben, tell us a little bit about your You're a little bit to our normal interviews because you're not from a startup, but you are from a small to medium enterprise. why don't tell us a bit about UK Coaching, what you do and like where the idea came the app that you ⁓ then went on to build. Yeah, I I think I might blend in two here, but but really it's about they're ⁓ not our typical customer in terms of they're a bit bigger, they're a startup, ⁓ they're an established business. but the way that they approached ⁓ the kind of project, ⁓ one was unique because there was none of that corporate Yeah, exactly. Welcome back everybody. took a look through some of the questions that we've had over the last couple of months ⁓ we will do a different or separate Ask Us Anything deep dive. But today we're just going to tackle these two questions. So what have we got,
Ben Dickens: Yeah, so UK Coaching is a charity and we are funded by Sport England to provide services to coaches across the UK. One our campaigns that we run is called Play Their Way and that's for child first We've been looking for...
Verena: obviously I want ⁓ start with a question that we got from Germany. ⁓ this is from Linda. And Linda asked, if you started Medialabs from scratch, what would you do differently because of AI?
Simon: elements to it. I mean, course there was, but but they had adopted the kind of the startup way of building things, which I think is really powerful. And and they were able to build something like lightning fast as a result there. But did all their homework. They spoke to customers. They did the customer interviews. They spent quite a lot of time get getting that data and ⁓ knowing what they wanted in the app from day one. And that massively helps us and them because it means the output outputted product is much much closer to what they thought it was going to be. and you know were able to deliver really quickly for them. obviously.
Ben Dickens: Any sort of tools that coaches can use to help them in their coaching practice. one of the ideas, particularly on the Play Their Way campaign ⁓ to be able to produce a tool that would enable a coach ⁓ get child first coaching content. We getting a coach in your pocket via an app was probably the best way to do that. So we on a development journey to develop an app with sprinkling of AI,
Simon: So great question, Linda. So what would we do if we started this year from scratch? it's a really good question. mean, AI has changed everything. So we probably wouldn't have built out as big a team as we've built out, ⁓ I'm guessing, because we would have able to do a lot more with AI. ⁓
Verena: So it sounds like startup spirit was matched corporate. ⁓
Simon: I think Uber could have started without drivers, they would have done. Like you just would have, it just makes business sense. Not taking into account all of the moral and, ⁓ you know, and all those other bigger issues. But just from business point of view, like you just started out with less people if you could have done. you know, we would have probably, as ⁓ you might say, is we have built out more tools to people on the full journey.
Ben Dickens: would entice to download the app, ask questions of it and to our knowledge base of ⁓ questions being asked and producing content to support the coaches while they're doing their coaching or after sessions and before sessions.
Simon: and this is coaches across the board. this might be, ⁓ know, football rugby, ⁓ any sport is a coach that's coaching children. So that start, launch and grow. You know, we were very, to start with, we very focused on the launch, like just how do you build an app? And actually ⁓ just one element of what you need to do to get to kind of from idea to revenue. So I think we would have built out more.
Ben Dickens: Yep, essentially, so multi-skills, multi-sports, football, cricket, hockey, gymnastics, all the big sports in the UK. It's not specific to any sport. There is content that we've built into the app that is sport specific. So some of the partners that we work with, there's hockey. cricket, football, boxing, BMX, cycling, any sport where you are coached essentially, we've content in the base for the app to use ⁓ and refer to build its answers. It's quite a wide variety of data. There's some sensitive subjects in there like safeguarding, protecting different content in there that's been moderated, cleansed, ⁓ chunked up, ready for within the app.
Verena: yes, have one thought that's similar to yours then one thought that is not really related to the question. you're right. I would say we about to start again, I would make idea actually figuring what actually makes sense to build a core part of our offering. do that today, but it's the hardest part. And ⁓ one of the important part actually AI can't replace for you. ⁓ So focus more on and also the whole go-to-market. So what you mentioned in terms of to grow, Is that thing key? No one really wants to hear it.
Simon: And in terms coming up with the idea and the kind of the initial concept of what we would call the scope for the did that come from? Because you've got lots of stakeholders in there. So how did that kind of come about?
Ben Dickens: So worked with various agencies and partners, to, develop what our strategies over the course on that, and usually on a three year cycle. once we'd got to the point of that, we didn't really know what the app wanted and million labs were very good in, ⁓ helping us. they, termed that.
Verena: And this is, think, no matter whether you are a startup founder who needs to get paying customers or you're an innovator within an established organization who needs to justify the spending, And you need to get a return on investment. You also need this ⁓ validation of what you actually want to build, et cetera. yeah, think that having this service really also focused around that ⁓ of the I think, is definitely critical and becomes even more critical.
Ben Dickens: concept of a sprinkling of AI. So very topical at the moment. It's something that you can put into the ⁓ that's a hook to get people to come in and come and use it again. So it was combination of support from third parties, our own internal about what coaches wanted. We did great piece discovery work early doors
Simon: as you say, it's one of the most ⁓ underestimated bits lots of startup founders don't do. But I feel like we're banging this drum and we're constantly banging, banging, banging the same drum. So we're going to get boring. But yeah, ⁓ like you've to do your homework. You've got to do.
Ben Dickens: with coaches that said, you know, what are you looking for? What would be useful? What would help you on a day to day? we started to build a picture of what this app might look So that the research we did was firstly of whether an app would be useful. And a lot of coaches said, absolutely. There's nothing particularly out there at the moment that is doing this and certainly not the content around child first coaching. So it supported campaign really well. it us to.
Verena: Maybe Simon, just have to realize that people just don't want to listen to us. Maybe it's a Simon Verena problem, not a market problem. I don't know.
Simon: Yeah, that's it. Well, I think they want to get fit and healthy and lose weight by just taking the jab and not by doing any exercise or changing their diet. It's the same principle, isn't it? We're humans after all. We don't want to do any of the hard stuff.
Ben Dickens: bring that new technology into a campaign and launch something, you know, quite early in comparison to a lot of the campaigns at the moment in terms of certainly Child First certainly within the sports sector as you know, we're quite proud the work that we've done and where we've got to at this stage in terms of building something that works really nicely.
Verena: And yeah, I I mean, ⁓ that wasn't Linda's but Linda, so if you were to ask like what we would do the same. ⁓ So I think what very well is we ⁓ didn't tie our vision and our position ⁓ to a specific technology. So we started back then, ⁓ we focused a lot on no code, it's more obviously AI, right? And five years from now,
Simon: And love that you talked about talking to coach. We call it customer interviews. ⁓ So startups don't do that. They don't and talk to the actual customer that will be using it, but you've done that. certainly for us that having input from ⁓ the customer or, know, or the user of this, they're not paying for but you the user of that is, like, is worth a million dollars because you could really then start to know which features are important, which features may be less important,
Verena: it's probably something completely different, right? So our core mission of launching one within startup state unaffected. And I think this helped us a lot. And so I think something that regardless of the whole AI shift is what we would do the same.
Simon: As we all know, the only way to really find out is get it in people's hands is, you know, is getting, it out there. you through the launch process with us terms of the build, et cetera, I suspect, certainly in talking to you and your team that sort of flow we do, cause we work at, I guess, startup speed ⁓ was probably a little bit different to maybe how you've dealt with the organizations before. Yeah, I think that's totally true. We've always seen like the technology is just a tool for getting from idea to revenue to kind of getting the startup to the next stage. yeah, not about pinning yourself to a specific technology. It's about, you using what's the best thing available to you at that point. So yeah, totally agree. question two. This one's from in the UK.
Ben Dickens: Yes, it was very different, very different. So we've done lots of website builds over the years for UK coaching for the Play Their Way And we've done website upgrades and migration paths moving CMSs. And the time it to those is, is fairly established. You know, talking, you know, a six to 12 month process of selecting the CMS, putting the content in building out, building your widgets and so on. So we were absolutely caught by One at the initial proposal that we
Simon: What's one AI prediction for the next 12 months?
Verena: I mean, I think for AI becomes really real, right? Like I think so far from what we see, also from talking with friends, et cetera, people have treated AI as something super important. And I think they understood that, yeah, this is big, right? But it still seemed to be a bit in the background. They use ChatGPT, they attended a few seminars, right, of how to use it for your organiSation, right? they watched videos, I think a lot of
Ben Dickens: could have a working app within two to three months. the reduced lead times have been able to start the project. Three, being able to go from a discovery phrase to a, to a functional specification, essentially, you know, a light touch one in terms of what we were going to do and how we're going to do it. and then the rapid speed of development, you know, we've not worked at that pace before and it catch you out in terms of ⁓ have got enough internal resource to ⁓ build for the library to do creative that we need to do that. And, and obviously Million Labs will give you this template where a whole heap of integrations
Verena: companies also got training, which is great. People started to experiment with Claude, but I think most people, haven't really been affected ⁓ by consequences. And I think over the next 12 months I think we will start seeing much more direct impact. So ⁓ which is company you are working at or your own company doing well, because you managed to to ⁓ the big shift.
Ben Dickens: you need because we're using cloud technologies. So creating accounts, setting those up, it was almost as much work to create all of the various accounts for, the tools that we use in, ⁓ the cloud than it was to actually develop the software to get us to where we needed to be. So it was an extremely quick process. it's us to iterate as well. So version one of the app, we're probably on.
Verena: Or are you now struggling and might actually disappear? We will see, I think a lot of these Kodak and Blockbusters, examples of companies that try to kind of defend the status quo. And you don't see that from day one. You don't see a shift happening in the first year, but you see of a sudden it's just there ⁓ And I think we are entering the where people stop talking about AI as a future trend,
Ben Dickens: what I would traditionally call probably version three now where we've had iterations throughout the process where we've tried, we've tested it, we've done a bit more discovery and research, we've shown it to people that we've come back. funders have come back with questions about, ⁓ are the of the AI? How do we counteract some of these questions, people may ask. And it's just been incredibly quick process. We have tied this in with our website as well. So our launch campaigns are...
Verena: and start experiencing it in their daily work and life. And the question is no longer whether AI will change things, right? But rather, of us decides to do with a change and how it personally affects us.
Ben Dickens: multifaceted. and one of the big pieces of work with the whole, the whole project really was to use, single sign on tool so that we had a central source of truth for our customer. So our website could log in with their website, account into the app and our ⁓ customers can log the website with their, the same account. So we've got that single source of truth. So we can see what content is being consumed in the app and what content is being consumed in the website app for the same user at the same time, essentially.
Simon: So I think my take is probably a bit around what's happening. ⁓ And once obviously a technology point of view, it's moving fast. We like literally models every, ⁓ every couple months, new versions of the models and they get better and better.
Ben Dickens: So to get that far the multi components, it's really and really quick.
Simon: you know, that's a positive thing, I think, for startup founders, because you can do more quicker ⁓ get better results. I think the misunderstood piece around that is what we call the false that come out of it. So Mythos apparently is currently running in like 30 % false positives, which means that if you ask it, questions, 30 of the time it's going to get it wrong, or it's going to say, you know, this is not when it actually is working. that's not us being geniuses. Lots has in the market. AI ⁓ is allowing us to able to move quick and to customers quick. And so
Ben Dickens: Yeah. Yeah. I think that ultimately the tools are available to enable this to happen at the point in time that it happened. And even, even up to the point where we were starting work, we were still thinking about what, what tools come under most of the past six months. What, what can we use here? What can we use there? and again, a lot of the, software, cloud software that we've put together.
Simon: And that's particularly an issue when you're looking at coding or security or things like that. Cause you, you know, you're going to waste an extra 30 % of the time proving that actually mythos was wrong. It wasn't the security bug was actually just correct at the way the code works. I think that's going to be one challenge that not very understood to, think we're still towards the multi-agent world.
Ben Dickens: Even as we've been building this project, they've been releasing new updates within the cloud and given us new functions as we're building. And we've been able to react and put something brand new in within hours, not even days. ⁓ let's try this. Let's try that. This new functionality exists. Can we piggyback off the back of that? And we have been able to do that.
Simon: Lots of wrappers are kind of still in that single agent chat to an agent phase, now we're starting to see startups to use multi-agents. So you have one for doing design, one for doing coding or one, you know, whatever, whatever your use case is. ⁓ So think we're going to see a lot, lot more of that. And then I think looking a bit further out, we're going to see deep integrations into product. And at that point, becomes absolutely critical. brilliant and at the same time because things are moving so quickly. you start building with one version of ChatGPT or Anthropical whatever it is. the time we get to the end, maybe another one or two versions out that actually are just better. So, build complete, launch. I think you've soft launched. because you want as a business, you want it working on your data. So if you're collecting data from clients, you know, take us, we're working with lots of startups. We want to be able to ask our agent a question about our startups, like which startup bought this product or which startup is at this stage. And we just ask it a question. that ⁓ that it has to have access to our data, obviously securely. So think that's going to become a really big thing for all businesses over the next 12 months. So you've started to kind of push it out to various customers, but you've got the real launch of the app later next month or something like that. So how's it been received? Like how's the kind of first soft launch gone?
Ben Dickens: Yeah, soft launch has gone fine. So of the tricky bits around single sign on, ⁓ sure that our web accounts were shared with the app. that's probably been more as much work in getting that work, know, as in actually building the app and the AI piece. we're still adding still content. We're learning how to chunk data up for AI. done some work alongside the project Million Apps to allow us to scrape.
Verena: So now we covered the personal, business and technical perspective. I think that's enough. ⁓ so as we said, guys, we will do a whole deep dive about questions, if you have something that you want us to answer, just send us your topic at hello at millionlabs.co.uk. We will also add the email to the show notes. But for today, I would say let's...
Simon: There you go. Tick, tick, tick and tick.
Ben Dickens: websites to take content out from partners that have approved it. then we're going to do a bit more work in the future to be able to automatically chunk that and keep that topped up. really on the ⁓ app to sure our content is moderated and clean and So we're not scraping the internet for stuff that we don't know. It's content that's in closed loop that's either our own UK coaching or Play the Waste content ⁓ or for the partners we work with.
Verena: Let's welcome today's guest. Let's jump into the interview with
Ben Dickens: So he's taken a bit, a bit of time to do that in terms of the content piece. but it's ready for launch. We've got there, we've got two to 300 pieces of content in there, fairly lengthy articles and guidance for coaches. the administration portal to allow us to manage our own content, manage users, check settings, do the notifications for the app. He's up and running alongside it as well. so we'll be in a position to full full launch very soon. ⁓ as I say, it's in both app stores, Apple and Google. the moment it supports single sign-on you can download it now ⁓ essentially it works
Simon: How you go about thinking about some of the around the AI? What did you do with the development team to ⁓ just about that?
Ben Dickens: got some good internal knowledge. We've our safeguarding lead, Heather, we've got lots of experience within the organisation of dealing with sensitive subjects, particularly around safeguarding and things like that. You know, that's, that's part of our sector. You know, UK coaching runs lots of safeguarding e-learning and workshops. So we've got lots of expertise in the business. We did lots of internal testing and checking of these questions and making sure that. any sensitive questions were asked, we referred to the right content. certainly put, we ring-fenced some of statements in our way to make sure if there was anything that cropped up that we couldn't answer, we made sure it was clear, we put disclaimers in We also host quite bit of NSPCC within the knowledge bank as well in terms of good So across the board, you know, good testing. disclaimers in place and making sure that your AI rules are in place to say, well, if somebody does ask a question and we're not quite sure, don't make the answer up. also reference for every AI question asked. also reference it with these are the five, six, seven articles that this content has come from. So in the event of a question that the coach asks, and we, don't know the answer somebody we don't think this answer is right. We know where the source of the data is. can get, we can route straight down to that particular article and say, actually, ⁓ you know, article one, two, three, four from UK coaching. There's lots of evidence and research to this up to actually. is a good answer or this is a coach. But again, it's work in progress. We're always going to be maintaining this content base of, protections, and making sure we come up with the right answer when coaches ask it. We're also storing coaches questions in a separate database. it's helping us in our content strategy to be able to say, well, what are the questions coaches are asking that we can't answer? And then building out from there and putting content in to support the questions that are being answered well or being not answered at all.
Simon: I think people get worried when start saying AI and questions and so forth. But actually, if put the safeguards around it or the control measures around it to say, only look up from our content and only respond from our content, et cetera, it can't go off to the internet and find something and make something up, et cetera, and try and cut down those hallucinations, which to be fair, getting much, much fewer then yeah, I think you can get some good results.
Ben Dickens: And we've got our own safeguards in place. We've got a good governance team. We've gone through the process. We've done our agreements. We've updated our policies. We've got an AI policy. We've done all that compliance work to make sure this is robust, stands its own, and is essentially. So a ⁓ lot of work has gone into that.
Simon: Yeah. if there's another out there who are thinking of sort of going down this path of iterating an app, building an et cetera, whether that be a web app or a mobile app, that happens to be, I guess, what's your learning? What's your advice to them? Like worked? What didn't work? What would you change? What would you do differently?
Ben Dickens: it's quite difficult to find the right partners to work with, particularly the app side, we hadn't had any experience of working on applications before, mobile apps before. We've done lots and lots of website ⁓ with increased speed comes cost as well. So ⁓ prototyping is key. Getting first draft of something out there that you can see and test and work. ⁓ You'll know very quickly a matter of weeks, whether We're going down the right path, whether things are working, whether we're going to be on track. So I think with longer projects, particularly on websites, you know, you don't really know where you are for the first month or two, because you haven't, you don't get to see anything for quite a period until actually rolled out. Whereas million labs in terms of the app, we had something that we could see in its first version within certainly less than three or four weeks that we could then start testing and have a look at, and then go from there and build on that. So it was, that's what sped up the process as well.
Simon: And I that's a change that's coming to, you know, all outsourced agency developers, et cetera. It's that ability to better get something in front of the customer really quickly. Cause that's absolutely critical. Otherwise, you know, you wanted a four bedroom house and what ended up with is was a shed, you know, it's, it's the paths diverge and you don't quite get what you thought were going to get. and as you say, the longer the project goes on that, that to keep those expectations in line,
Ben Dickens: changes your, um, your budgeting process a little bit as well, because rather than just sort of that, that fixed price at the end of the day, this is what you're going to get. can, you can intercept that in, in three to four or five weeks and say, Oh, actually, well, I didn't know we could do this. Let's, let's try this bit and see where that goes. and that's worked really well for us because we have been able to add value throughout the process. What is really surprising, what a lot of people have said to us as well is the first version released to the app store looks so polished, it looks like it's been around for months and months, if not years. Whereas, whereas in the past you build something that, you know, there's a few bits of work you've got to do or you've got to finish. Whereas this looks incredibly polished from day You know, the color schemes, the graphics, it's completely in line with our brand guidelines the functionality is really rich as well. you ⁓ know, digital badges. Traditionally, that's something we would probably add into a piece of software further down the road once we've got the core functionality working. But we've been able to add all the bells and whistles from day one, which is unusual, certainly.
Simon: also the ongoing ability to iterate at speed is critical because now as the coaches actually start using it for real and live, you can take on board that feedback and you can be making changes like, know, within, within days. Okay. Well, really appreciate you sitting down with us and sharing your, your journey. I, hopefully the full launch goes well. If people are coaches out there, because I'm sure there are lots of coaches here in the UK, what are they looking for in the App Store? What are they searching for?
Ben Dickens: It's called Play Their Way, simple as that. So coaching your pocket so they can go to the playtheirway.org website. and it will be promoted heavily on there in the coming days.
Simon: brilliant. All right, Ben, really appreciate your time.
Simon: Ben, tell us a little bit about your You're a little bit to our normal interviews because you're not from a startup, but you are from a small to medium enterprise. why don't tell us a bit about UK Coaching, what you do and like where the idea came the app that you ⁓ then went on to build. Yeah, I I think I might blend in two here, but but really it's about they're ⁓ not our typical customer in terms of they're a bit bigger, they're a startup, ⁓ they're an established business. but the way that they approached ⁓ the kind of project, ⁓ one was unique because there was none of that corporate Yeah, exactly. Welcome back everybody. took a look through some of the questions that we've had over the last couple of months ⁓ we will do a different or separate Ask Us Anything deep dive. But today we're just going to tackle these two questions. So what have we got,
Ben Dickens: Yeah, so UK Coaching is a charity and we are funded by Sport England to provide services to coaches across the UK. One our campaigns that we run is called Play Their Way and that's for child first We've been looking for...
Verena: obviously I want ⁓ start with a question that we got from Germany. ⁓ this is from Linda. And Linda asked, if you started Medialabs from scratch, what would you do differently because of AI?
Simon: elements to it. I mean, course there was, but but they had adopted the kind of the startup way of building things, which I think is really powerful. And and they were able to build something like lightning fast as a result there. But did all their homework. They spoke to customers. They did the customer interviews. They spent quite a lot of time get getting that data and ⁓ knowing what they wanted in the app from day one. And that massively helps us and them because it means the output outputted product is much much closer to what they thought it was going to be. and you know were able to deliver really quickly for them. obviously.
Ben Dickens: Any sort of tools that coaches can use to help them in their coaching practice. one of the ideas, particularly on the Play Their Way campaign ⁓ to be able to produce a tool that would enable a coach ⁓ get child first coaching content. We getting a coach in your pocket via an app was probably the best way to do that. So we on a development journey to develop an app with sprinkling of AI,
Simon: So great question, Linda. So what would we do if we started this year from scratch? it's a really good question. mean, AI has changed everything. So we probably wouldn't have built out as big a team as we've built out, ⁓ I'm guessing, because we would have able to do a lot more with AI. ⁓
Verena: So it sounds like startup spirit was matched corporate. ⁓
Simon: I think Uber could have started without drivers, they would have done. Like you just would have, it just makes business sense. Not taking into account all of the moral and, ⁓ you know, and all those other bigger issues. But just from business point of view, like you just started out with less people if you could have done. you know, we would have probably, as ⁓ you might say, is we have built out more tools to people on the full journey.
Ben Dickens: would entice to download the app, ask questions of it and to our knowledge base of ⁓ questions being asked and producing content to support the coaches while they're doing their coaching or after sessions and before sessions.
Simon: and this is coaches across the board. this might be, ⁓ know, football rugby, ⁓ any sport is a coach that's coaching children. So that start, launch and grow. You know, we were very, to start with, we very focused on the launch, like just how do you build an app? And actually ⁓ just one element of what you need to do to get to kind of from idea to revenue. So I think we would have built out more.
Ben Dickens: Yep, essentially, so multi-skills, multi-sports, football, cricket, hockey, gymnastics, all the big sports in the UK. It's not specific to any sport. There is content that we've built into the app that is sport specific. So some of the partners that we work with, there's hockey. cricket, football, boxing, BMX, cycling, any sport where you are coached essentially, we've content in the base for the app to use ⁓ and refer to build its answers. It's quite a wide variety of data. There's some sensitive subjects in there like safeguarding, protecting different content in there that's been moderated, cleansed, ⁓ chunked up, ready for within the app.
Verena: yes, have one thought that's similar to yours then one thought that is not really related to the question. you're right. I would say we about to start again, I would make idea actually figuring what actually makes sense to build a core part of our offering. do that today, but it's the hardest part. And ⁓ one of the important part actually AI can't replace for you. ⁓ So focus more on and also the whole go-to-market. So what you mentioned in terms of to grow, Is that thing key? No one really wants to hear it.
Simon: And in terms coming up with the idea and the kind of the initial concept of what we would call the scope for the did that come from? Because you've got lots of stakeholders in there. So how did that kind of come about?
Ben Dickens: So worked with various agencies and partners, to, develop what our strategies over the course on that, and usually on a three year cycle. once we'd got to the point of that, we didn't really know what the app wanted and million labs were very good in, ⁓ helping us. they, termed that.
Verena: And this is, think, no matter whether you are a startup founder who needs to get paying customers or you're an innovator within an established organization who needs to justify the spending, And you need to get a return on investment. You also need this ⁓ validation of what you actually want to build, et cetera. yeah, think that having this service really also focused around that ⁓ of the I think, is definitely critical and becomes even more critical.
Ben Dickens: concept of a sprinkling of AI. So very topical at the moment. It's something that you can put into the ⁓ that's a hook to get people to come in and come and use it again. So it was combination of support from third parties, our own internal about what coaches wanted. We did great piece discovery work early doors
Simon: as you say, it's one of the most ⁓ underestimated bits lots of startup founders don't do. But I feel like we're banging this drum and we're constantly banging, banging, banging the same drum. So we're going to get boring. But yeah, ⁓ like you've to do your homework. You've got to do.
Ben Dickens: with coaches that said, you know, what are you looking for? What would be useful? What would help you on a day to day? we started to build a picture of what this app might look So that the research we did was firstly of whether an app would be useful. And a lot of coaches said, absolutely. There's nothing particularly out there at the moment that is doing this and certainly not the content around child first coaching. So it supported campaign really well. it us to.
Verena: Maybe Simon, just have to realize that people just don't want to listen to us. Maybe it's a Simon Verena problem, not a market problem. I don't know.
Simon: Yeah, that's it. Well, I think they want to get fit and healthy and lose weight by just taking the jab and not by doing any exercise or changing their diet. It's the same principle, isn't it? We're humans after all. We don't want to do any of the hard stuff.
Ben Dickens: bring that new technology into a campaign and launch something, you know, quite early in comparison to a lot of the campaigns at the moment in terms of certainly Child First certainly within the sports sector as you know, we're quite proud the work that we've done and where we've got to at this stage in terms of building something that works really nicely.
Verena: And yeah, I I mean, ⁓ that wasn't Linda's but Linda, so if you were to ask like what we would do the same. ⁓ So I think what very well is we ⁓ didn't tie our vision and our position ⁓ to a specific technology. So we started back then, ⁓ we focused a lot on no code, it's more obviously AI, right? And five years from now,
Simon: And love that you talked about talking to coach. We call it customer interviews. ⁓ So startups don't do that. They don't and talk to the actual customer that will be using it, but you've done that. certainly for us that having input from ⁓ the customer or, know, or the user of this, they're not paying for but you the user of that is, like, is worth a million dollars because you could really then start to know which features are important, which features may be less important,
Verena: it's probably something completely different, right? So our core mission of launching one within startup state unaffected. And I think this helped us a lot. And so I think something that regardless of the whole AI shift is what we would do the same.
Simon: As we all know, the only way to really find out is get it in people's hands is, you know, is getting, it out there. you through the launch process with us terms of the build, et cetera, I suspect, certainly in talking to you and your team that sort of flow we do, cause we work at, I guess, startup speed ⁓ was probably a little bit different to maybe how you've dealt with the organizations before. Yeah, I think that's totally true. We've always seen like the technology is just a tool for getting from idea to revenue to kind of getting the startup to the next stage. yeah, not about pinning yourself to a specific technology. It's about, you using what's the best thing available to you at that point. So yeah, totally agree. question two. This one's from in the UK.
Ben Dickens: Yes, it was very different, very different. So we've done lots of website builds over the years for UK coaching for the Play Their Way And we've done website upgrades and migration paths moving CMSs. And the time it to those is, is fairly established. You know, talking, you know, a six to 12 month process of selecting the CMS, putting the content in building out, building your widgets and so on. So we were absolutely caught by One at the initial proposal that we
Simon: What's one AI prediction for the next 12 months?
Verena: I mean, I think for AI becomes really real, right? Like I think so far from what we see, also from talking with friends, et cetera, people have treated AI as something super important. And I think they understood that, yeah, this is big, right? But it still seemed to be a bit in the background. They use ChatGPT, they attended a few seminars, right, of how to use it for your organiSation, right? they watched videos, I think a lot of
Ben Dickens: could have a working app within two to three months. the reduced lead times have been able to start the project. Three, being able to go from a discovery phrase to a, to a functional specification, essentially, you know, a light touch one in terms of what we were going to do and how we're going to do it. and then the rapid speed of development, you know, we've not worked at that pace before and it catch you out in terms of ⁓ have got enough internal resource to ⁓ build for the library to do creative that we need to do that. And, and obviously Million Labs will give you this template where a whole heap of integrations
Verena: companies also got training, which is great. People started to experiment with Claude, but I think most people, haven't really been affected ⁓ by consequences. And I think over the next 12 months I think we will start seeing much more direct impact. So ⁓ which is company you are working at or your own company doing well, because you managed to to ⁓ the big shift.
Ben Dickens: you need because we're using cloud technologies. So creating accounts, setting those up, it was almost as much work to create all of the various accounts for, the tools that we use in, ⁓ the cloud than it was to actually develop the software to get us to where we needed to be. So it was an extremely quick process. it's us to iterate as well. So version one of the app, we're probably on.
Verena: Or are you now struggling and might actually disappear? We will see, I think a lot of these Kodak and Blockbusters, examples of companies that try to kind of defend the status quo. And you don't see that from day one. You don't see a shift happening in the first year, but you see of a sudden it's just there ⁓ And I think we are entering the where people stop talking about AI as a future trend,
Ben Dickens: what I would traditionally call probably version three now where we've had iterations throughout the process where we've tried, we've tested it, we've done a bit more discovery and research, we've shown it to people that we've come back. funders have come back with questions about, ⁓ are the of the AI? How do we counteract some of these questions, people may ask. And it's just been incredibly quick process. We have tied this in with our website as well. So our launch campaigns are...
Verena: and start experiencing it in their daily work and life. And the question is no longer whether AI will change things, right? But rather, of us decides to do with a change and how it personally affects us.
Ben Dickens: multifaceted. and one of the big pieces of work with the whole, the whole project really was to use, single sign on tool so that we had a central source of truth for our customer. So our website could log in with their website, account into the app and our ⁓ customers can log the website with their, the same account. So we've got that single source of truth. So we can see what content is being consumed in the app and what content is being consumed in the website app for the same user at the same time, essentially.
Simon: So I think my take is probably a bit around what's happening. ⁓ And once obviously a technology point of view, it's moving fast. We like literally models every, ⁓ every couple months, new versions of the models and they get better and better.
Ben Dickens: So to get that far the multi components, it's really and really quick.
Simon: you know, that's a positive thing, I think, for startup founders, because you can do more quicker ⁓ get better results. I think the misunderstood piece around that is what we call the false that come out of it. So Mythos apparently is currently running in like 30 % false positives, which means that if you ask it, questions, 30 of the time it's going to get it wrong, or it's going to say, you know, this is not when it actually is working. that's not us being geniuses. Lots has in the market. AI ⁓ is allowing us to able to move quick and to customers quick. And so
Ben Dickens: Yeah. Yeah. I think that ultimately the tools are available to enable this to happen at the point in time that it happened. And even, even up to the point where we were starting work, we were still thinking about what, what tools come under most of the past six months. What, what can we use here? What can we use there? and again, a lot of the, software, cloud software that we've put together.
Simon: And that's particularly an issue when you're looking at coding or security or things like that. Cause you, you know, you're going to waste an extra 30 % of the time proving that actually mythos was wrong. It wasn't the security bug was actually just correct at the way the code works. I think that's going to be one challenge that not very understood to, think we're still towards the multi-agent world.
Ben Dickens: Even as we've been building this project, they've been releasing new updates within the cloud and given us new functions as we're building. And we've been able to react and put something brand new in within hours, not even days. ⁓ let's try this. Let's try that. This new functionality exists. Can we piggyback off the back of that? And we have been able to do that.
Simon: Lots of wrappers are kind of still in that single agent chat to an agent phase, now we're starting to see startups to use multi-agents. So you have one for doing design, one for doing coding or one, you know, whatever, whatever your use case is. ⁓ So think we're going to see a lot, lot more of that. And then I think looking a bit further out, we're going to see deep integrations into product. And at that point, becomes absolutely critical. brilliant and at the same time because things are moving so quickly. you start building with one version of ChatGPT or Anthropical whatever it is. the time we get to the end, maybe another one or two versions out that actually are just better. So, build complete, launch. I think you've soft launched. because you want as a business, you want it working on your data. So if you're collecting data from clients, you know, take us, we're working with lots of startups. We want to be able to ask our agent a question about our startups, like which startup bought this product or which startup is at this stage. And we just ask it a question. that ⁓ that it has to have access to our data, obviously securely. So think that's going to become a really big thing for all businesses over the next 12 months. So you've started to kind of push it out to various customers, but you've got the real launch of the app later next month or something like that. So how's it been received? Like how's the kind of first soft launch gone?
Ben Dickens: Yeah, soft launch has gone fine. So of the tricky bits around single sign on, ⁓ sure that our web accounts were shared with the app. that's probably been more as much work in getting that work, know, as in actually building the app and the AI piece. we're still adding still content. We're learning how to chunk data up for AI. done some work alongside the project Million Apps to allow us to scrape.
Verena: So now we covered the personal, business and technical perspective. I think that's enough. ⁓ so as we said, guys, we will do a whole deep dive about questions, if you have something that you want us to answer, just send us your topic at hello at millionlabs.co.uk. We will also add the email to the show notes. But for today, I would say let's...
Simon: There you go. Tick, tick, tick and tick.
Ben Dickens: websites to take content out from partners that have approved it. then we're going to do a bit more work in the future to be able to automatically chunk that and keep that topped up. really on the ⁓ app to sure our content is moderated and clean and So we're not scraping the internet for stuff that we don't know. It's content that's in closed loop that's either our own UK coaching or Play the Waste content ⁓ or for the partners we work with.
Verena: Let's welcome today's guest. Let's jump into the interview with
Ben Dickens: So he's taken a bit, a bit of time to do that in terms of the content piece. but it's ready for launch. We've got there, we've got two to 300 pieces of content in there, fairly lengthy articles and guidance for coaches. the administration portal to allow us to manage our own content, manage users, check settings, do the notifications for the app. He's up and running alongside it as well. so we'll be in a position to full full launch very soon. ⁓ as I say, it's in both app stores, Apple and Google. the moment it supports single sign-on you can download it now ⁓ essentially it works
Simon: How you go about thinking about some of the around the AI? What did you do with the development team to ⁓ just about that?
Ben Dickens: got some good internal knowledge. We've our safeguarding lead, Heather, we've got lots of experience within the organisation of dealing with sensitive subjects, particularly around safeguarding and things like that. You know, that's, that's part of our sector. You know, UK coaching runs lots of safeguarding e-learning and workshops. So we've got lots of expertise in the business. We did lots of internal testing and checking of these questions and making sure that. any sensitive questions were asked, we referred to the right content. certainly put, we ring-fenced some of statements in our way to make sure if there was anything that cropped up that we couldn't answer, we made sure it was clear, we put disclaimers in We also host quite bit of NSPCC within the knowledge bank as well in terms of good So across the board, you know, good testing. disclaimers in place and making sure that your AI rules are in place to say, well, if somebody does ask a question and we're not quite sure, don't make the answer up. also reference for every AI question asked. also reference it with these are the five, six, seven articles that this content has come from. So in the event of a question that the coach asks, and we, don't know the answer somebody we don't think this answer is right. We know where the source of the data is. can get, we can route straight down to that particular article and say, actually, ⁓ you know, article one, two, three, four from UK coaching. There's lots of evidence and research to this up to actually. is a good answer or this is a coach. But again, it's work in progress. We're always going to be maintaining this content base of, protections, and making sure we come up with the right answer when coaches ask it. We're also storing coaches questions in a separate database. it's helping us in our content strategy to be able to say, well, what are the questions coaches are asking that we can't answer? And then building out from there and putting content in to support the questions that are being answered well or being not answered at all.
Simon: I think people get worried when start saying AI and questions and so forth. But actually, if put the safeguards around it or the control measures around it to say, only look up from our content and only respond from our content, et cetera, it can't go off to the internet and find something and make something up, et cetera, and try and cut down those hallucinations, which to be fair, getting much, much fewer then yeah, I think you can get some good results.
Ben Dickens: And we've got our own safeguards in place. We've got a good governance team. We've gone through the process. We've done our agreements. We've updated our policies. We've got an AI policy. We've done all that compliance work to make sure this is robust, stands its own, and is essentially. So a ⁓ lot of work has gone into that.
Simon: Yeah. if there's another out there who are thinking of sort of going down this path of iterating an app, building an et cetera, whether that be a web app or a mobile app, that happens to be, I guess, what's your learning? What's your advice to them? Like worked? What didn't work? What would you change? What would you do differently?
Ben Dickens: it's quite difficult to find the right partners to work with, particularly the app side, we hadn't had any experience of working on applications before, mobile apps before. We've done lots and lots of website ⁓ with increased speed comes cost as well. So ⁓ prototyping is key. Getting first draft of something out there that you can see and test and work. ⁓ You'll know very quickly a matter of weeks, whether We're going down the right path, whether things are working, whether we're going to be on track. So I think with longer projects, particularly on websites, you know, you don't really know where you are for the first month or two, because you haven't, you don't get to see anything for quite a period until actually rolled out. Whereas million labs in terms of the app, we had something that we could see in its first version within certainly less than three or four weeks that we could then start testing and have a look at, and then go from there and build on that. So it was, that's what sped up the process as well.
Simon: And I that's a change that's coming to, you know, all outsourced agency developers, et cetera. It's that ability to better get something in front of the customer really quickly. Cause that's absolutely critical. Otherwise, you know, you wanted a four bedroom house and what ended up with is was a shed, you know, it's, it's the paths diverge and you don't quite get what you thought were going to get. and as you say, the longer the project goes on that, that to keep those expectations in line,
Ben Dickens: changes your, um, your budgeting process a little bit as well, because rather than just sort of that, that fixed price at the end of the day, this is what you're going to get. can, you can intercept that in, in three to four or five weeks and say, Oh, actually, well, I didn't know we could do this. Let's, let's try this bit and see where that goes. and that's worked really well for us because we have been able to add value throughout the process. What is really surprising, what a lot of people have said to us as well is the first version released to the app store looks so polished, it looks like it's been around for months and months, if not years. Whereas, whereas in the past you build something that, you know, there's a few bits of work you've got to do or you've got to finish. Whereas this looks incredibly polished from day You know, the color schemes, the graphics, it's completely in line with our brand guidelines the functionality is really rich as well. you ⁓ know, digital badges. Traditionally, that's something we would probably add into a piece of software further down the road once we've got the core functionality working. But we've been able to add all the bells and whistles from day one, which is unusual, certainly.
Simon: also the ongoing ability to iterate at speed is critical because now as the coaches actually start using it for real and live, you can take on board that feedback and you can be making changes like, know, within, within days. Okay. Well, really appreciate you sitting down with us and sharing your, your journey. I, hopefully the full launch goes well. If people are coaches out there, because I'm sure there are lots of coaches here in the UK, what are they looking for in the App Store? What are they searching for?
Ben Dickens: It's called Play Their Way, simple as that. So coaching your pocket so they can go to the playtheirway.org website. and it will be promoted heavily on there in the coming days.
Simon: brilliant. All right, Ben, really appreciate your time.
