Mel Morris, CBE, interview with David Plumb. का पूरा ट्रांसक्रिप्ट

00:04:22.879 --> 00:04:29.439 Came in one day, and I'm upstairs in this meeting, and we're debugging a problem with a computer system. 00:04:30.079 --> 00:04:33.920 And the people in the room, most hadn't really worked with him. 00:04:34.240 --> 00:04:36.480 But he's asking, What problem? 00:04:36.720 --> 00:04:37.439 What problem? 00:04:37.920 --> 00:04:41.120 So one guy says, Don't worry, doctor, we'll figure it out. 00:04:41.199 --> 00:04:41.920 He says, What problem? 00:04:42.079 --> 00:04:43.040 I said, excuse me. 00:04:43.279 --> 00:04:45.759 Please tell him what the situation is. 00:04:46.000 --> 00:04:50.639 So he walks around the room with all these circuit diagrams, never seen them before in his life. 00:04:50.959 --> 00:04:55.279 He walks around, he gets his cane out, he slaps it down, goes, problem here. 00:04:55.680 --> 00:04:58.000 Fetch board and soldering iron. 00:04:58.480 --> 00:05:04.480 And he figured what this minute problem was that people have been working on for months. 00:05:04.800 --> 00:05:06.560 And I just thought, wow. 00:05:07.040 --> 00:05:08.399 And that set me on the path. 00:05:08.480 --> 00:05:12.000 It said, no matter how good you become, you're never going to be as good as him. 00:05:12.240 --> 00:05:17.279 And that inspired me to do all sorts of things in pursuit of trying to at least be better than I was. 00:05:17.600 --> 00:05:24.399 I worked for Wang doing computer tuning, going down the biggest customers, helping them get those, and then built a business doing that. 00:05:24.560 --> 00:05:26.959 After that was a business called Minitech. 00:05:27.199 --> 00:05:33.519 Then that then became ProMetrics, doing PC software, and then that got bought out by Platinum. 00:05:33.600 --> 00:05:35.439 I became New Product Development Manager. 00:05:35.600 --> 00:05:36.639 That sold. 00:05:37.120 --> 00:05:38.720 I then went to doing New Date. 00:05:41.680 --> 00:05:51.519 After that, I got involved in four other companies, one in social media listening, one in Google Advertising and those sorts of markets. 00:05:52.000 --> 00:05:58.720 One in terms of intelligent sensors, one in terms of today, which is Corpora. 00:05:59.199 --> 00:06:02.319 And so many businesses en route to that. 00:06:02.639 --> 00:06:10.079 In UD, two of the people who worked for me there, one was a guy called Toby Rowland, one was a guy called Ricardo Zacconi. 00:06:10.639 --> 00:06:12.480 Became close friends to both. 00:06:12.639 --> 00:06:17.439 They wanted to buy this small star to be in Sweden doing computer games. 00:06:17.680 --> 00:06:19.040 Computer games. 00:06:20.879 --> 00:06:21.680 I didn't join them. 00:06:21.839 --> 00:06:22.879 They said, Would you be chairman? 00:06:23.040 --> 00:06:24.560 I said, Yeah, would you help us raise money? 00:06:24.720 --> 00:06:25.199 Yes. 00:06:25.519 --> 00:06:30.560 We scoured Europe looking for funders to back what became king. 00:06:30.879 --> 00:06:31.600 No takers. 00:06:31.920 --> 00:06:33.680 For several years, no-takers. 00:06:33.920 --> 00:06:36.720 Until one point, they're about out of cash and they said, What are we going to do? 00:06:36.800 --> 00:06:38.959 I said, I suppose I've better write a check. 00:06:39.120 --> 00:06:46.800 So I wrote a check to Keith Monday, and I stayed as the chairman of that company until it went public on the MISE in 2014. 00:06:47.600 --> 00:06:53.759 And Candy Crush was about probably 10 years into the creation of that business. 00:06:54.079 --> 00:06:57.439 So, but we knew it was going to be a success from day one. 00:06:57.600 --> 00:06:59.759 Lots of things to talk about at some other time. 00:07:00.079 --> 00:07:01.519 But that was the journey. 00:07:01.680 --> 00:07:04.800 And but the King story isn't about me. 00:07:04.959 --> 00:07:06.879 I provide the finance, I chaired it. 00:07:07.040 --> 00:07:10.079 It was about the team that worked in that business. 00:07:10.240 --> 00:07:12.800 Ricardo was a fantastic CEO. 00:07:13.040 --> 00:07:16.560 Sebastian Knutzen was the head of design for what went on. 00:07:16.800 --> 00:07:21.279 I just had the pleasure of working with a great team and helping backing them in any way I could. 00:07:21.439 --> 00:07:23.120 Wow, that is quite a journey, isn't it? 00:07:23.199 --> 00:07:25.920 And um, you know, there's some really interesting things to pick out there. 00:07:26.000 --> 00:07:30.399 I mean, the giving up your job on the basis of hoping that you're gonna get it's gonna work after three. 00:07:30.639 --> 00:07:32.879 I mean, that is really bold and brave. 00:07:33.040 --> 00:07:37.920 Um, and I love the humility, you know, the bringing together the people, it is fantastic. 00:07:38.079 --> 00:07:42.319 Um Candy Crush might be the one where I'm a customer, I've got to be honest. 00:07:42.480 --> 00:07:45.920 Um I'm not sure I did. 00:07:50.240 --> 00:07:51.920 It's too late now, I'm afraid. 00:07:52.079 --> 00:08:01.759 Um, but actually, interestingly, it was known as an incredibly innovative business, and I think it'd be interesting if you if it's okay to perhaps bring out why it was so innovative. 00:08:02.160 --> 00:08:09.680 Yeah, um, I mean the game was important, but it's not, in my opinion, the major part of that story. 00:08:09.920 --> 00:08:15.199 The game was important because it was a game that could be played by anyone pretty much of any age. 00:08:15.360 --> 00:08:19.040 So it was ubiquitous in its in its market opportunity. 00:08:19.519 --> 00:08:30.959 But but the secret, in my opinion, was again coming back to Ricardo and the development team, and and we'd figured out that we were doing computer games that were effectively tournament games. 00:08:31.120 --> 00:08:36.240 So you could sign up to play online pool against other people, and it would be a tournament. 00:08:36.320 --> 00:08:40.639 So lots of people would enter the game, sometimes two, three, four thousand people. 00:08:40.799 --> 00:08:50.960 They'd all pay a

few euros to play to get to one prize that one person would win the tournament, and that one person would probably walk away with the check for three, four thousand euros. 00:08:51.679 --> 00:08:55.600 But maybe three thousand people played that game to get to that point. 00:08:56.240 --> 00:09:00.480 How many people wanted to talk about it after the tournament had finished, do you think? 00:09:00.720 --> 00:09:04.639 Do you think it was 3,000 or do you think it was one person who walked away with the check? 00:09:05.360 --> 00:09:08.320 And and we figured out that you can't scale. 00:09:08.639 --> 00:09:10.159 You can't scale that. 00:09:10.399 --> 00:09:11.600 You can't scale that. 00:09:11.679 --> 00:09:13.519 There's no viral link to that. 00:09:13.679 --> 00:09:18.799 Because 2,990 people are absolutely fed up because they're wasted their money. 00:09:19.039 --> 00:09:22.399 Particularly those that got closest to the finish line. 00:09:22.720 --> 00:09:29.039 So we the business model was fantastic, it moved from being a tournament. 00:09:29.200 --> 00:09:30.320 Think of it as a cup. 00:09:30.480 --> 00:09:34.559 Those of you into soccer or rugby will know on a cup, it's a quick death. 00:09:34.960 --> 00:09:38.480 You you go in a round, you lose, you don't go to the next round. 00:09:38.799 --> 00:09:41.200 So on a cup, there's one winner. 00:09:41.519 --> 00:09:45.679 But you think about the Premier League, and it's a league. 00:09:46.000 --> 00:09:48.240 You come number two, it's a pretty good place to be. 00:09:48.399 --> 00:09:52.159 Actually, if you come number 19, it's a pretty good place to be. 00:09:52.320 --> 00:09:53.679 Even better if you're 16. 00:09:53.919 --> 00:09:58.000 So it's one of those things where we wanted a league. 00:09:58.240 --> 00:10:05.600 And then we looked at companies like FIFA, who had the brilliant model, which was well, you want to play FIFA, which one are you on? 00:10:05.759 --> 00:10:09.440 I'm on 2026, so I've only got 2023 still, it's not as good. 00:10:09.519 --> 00:10:12.879 It doesn't have all the new players, the new teams in, and so on and so forth. 00:10:13.120 --> 00:10:17.120 So we wanted a league, but we wanted a perennial game. 00:10:17.360 --> 00:10:19.200 So, what do you call a perennial game? 00:10:19.360 --> 00:10:20.799 You call it a saga. 00:10:21.200 --> 00:10:23.519 So can you crush became a saga? 00:10:23.600 --> 00:10:25.039 It's never going to end. 00:10:25.200 --> 00:10:28.159 It's going to have more levels than you could ever count. 00:10:28.480 --> 00:10:30.720 And where does the league come from? 00:10:30.879 --> 00:10:33.039 It's pretty easy, it's going to be your friends. 00:10:33.200 --> 00:10:38.879 And because everyone can play it, people will begin taunting one another on what level they've got to. 00:10:39.120 --> 00:10:43.840 So the model was you can play the game, two, three, four lives. 00:10:44.080 --> 00:10:54.240 When you run out of lives, you can sit and wait for

an eternity while your friends are getting ahead, or you can pay, or you can refer n number of your friends to get more lives. 00:10:54.720 --> 00:10:58.480 It was the ultimate expression of a viral game. 00:10:58.879 --> 00:11:02.720 The business model made the game, it made the company. 00:11:02.960 --> 00:11:10.240 And at peak, we almost got to ten million dollars per day in revenue. 00:11:11.039 --> 00:11:14.639 But only around three or four percent of people ever paid. 00:11:14.720 --> 00:11:15.679 David didn't. 00:11:17.200 --> 00:11:18.080 It's disgusting, really. 00:11:18.399 --> 00:11:19.360 I think my kids did. 00:11:20.639 --> 00:11:28.720 So so it was the game had to be something that anyone could play, from grandparents right the way down to grandkids. 00:11:28.960 --> 00:11:30.320 They could taunt each other. 00:11:30.480 --> 00:11:31.600 Where are you at, grandma? 00:11:31.679 --> 00:11:32.879 Oh, yeah, how many of this? 00:11:33.039 --> 00:11:35.519 Okay, I'm at 4,022. 00:11:35.919 --> 00:11:44.320 And and the frustration of having to wait got so bad that I know David actually got his credit card out on several occasions to get ahead. 00:11:44.480 --> 00:11:46.960 All right, and that's what happened, and the rest was history. 00:11:47.120 --> 00:11:48.639 Is that a good answer? 00:11:48.799 --> 00:11:50.320 It it's an amazing answer, isn't it? 00:11:50.480 --> 00:11:55.840 I mean, you think about some of the things that have happened in the last decade, you know, the Roblox and the Fortnites. 00:11:55.919 --> 00:12:13.759 Um, you started all of that in terms of them thinking about it as an ecosystem, thinking about it as member get member, community, uh, the freemium model, getting people hooked involved, sorry, and then and then figuring out how to, you know, it really was so innovative, um, which is incredible. 00:12:13.840 --> 00:12:23.440 And kind of over this last decade, we've seen a lot of changes with the tech, and you know, I can't do a full sentence without somebody mentioning AI to me at the moment. 00:12:23.600 --> 00:12:25.679 Um, it's very difficult to get through. 00:12:26.000 --> 00:12:32.240 I'm just interested in your perspective in terms of AI, the challenges, the opportunities, and where you see that going. 00:12:32.639 --> 00:12:34.480 Okay, just a quick show of hands. 00:12:34.799 --> 00:12:39.039 Has anyone ever heard of a paper called Attention is Everything? 00:12:39.679 --> 00:12:41.120 Okay, all right. 00:12:41.360 --> 00:12:43.519 So that paper, and you should answer that. 00:12:43.600 --> 00:12:45.519 If you if he can't, he should be fired. 00:12:45.759 --> 00:12:46.080 Okay. 00:12:47.120 --> 00:12:49.759 Otherwise, you're you're a complete facade, otherwise. 00:12:49.840 --> 00:12:55.519 So that paper was was authored by about four or five people in Google at the time. 00:12:55.679 --> 00:13:00.159 Uh, one of the guys, a guy called Leon Jones, uh, younger cunz, a couple of others as well. 00:13:00.399

--> 00:13:04.720 And that was the paper that basically figured out transformers. 00:13:04.879 --> 00:13:08.480 I don't mean the things, the robots that turn into whatever cars and things. 00:13:08.639 --> 00:13:19.360 It was the transformers that became the model of how we can actually predict the next word in sequence of a body of text, of a, of a, of a topic. 00:13:19.840 --> 00:13:24.480 And we saw that paper and we looked at it, thought, wow, this is really interesting. 00:13:24.559 --> 00:13:25.840 We went on a different path. 00:13:26.080 --> 00:13:34.240 But that paper was absolutely the starting point for today's incarnation of the tools we see today, ChatGPT. 00:13:34.480 --> 00:13:39.759 AI, its origins go way before that, but that was the trigger point. 00:13:40.159 --> 00:13:46.639 Then ChatGPT, October, November 2023, you know, they they launched. 00:13:46.720 --> 00:13:50.639 I think they're launched in some ways too soon, but it's typical of American companies. 00:13:50.799 --> 00:13:52.159 Let's grab the market. 00:13:52.320 --> 00:13:57.360 It may not be fully baked, it may not be quite ready, but we'll get it out there, we'll get a footprint. 00:13:57.519 --> 00:14:02.159 We put loads of money about promoting it, and by gum, did it take off? 00:14:02.399 --> 00:14:03.360 Absolutely took off. 00:14:03.440 --> 00:14:12.799 I mean, even Eclipse candy crushes growth in people terms, and and they were able then to get their valuation up, which let them borrow more money and away they would go. 00:14:13.039 --> 00:14:25.279 And Google did the typical thing that you see in these things when someone does something that causes, I think it was a half a trillion of value, shifted around the market in the space of a day or so. 00:14:25.519 --> 00:14:30.080 It shifted the market, it changed the whole dynamics of it. 00:14:30.240 --> 00:14:35.200 So Google is saying, hey, our stock price just tanked because these guys have launched this. 00:14:35.440 --> 00:14:37.600 But we've been working on that, we were the founders of this thing. 00:14:37.679 --> 00:14:38.480 What are we going to do? 00:14:38.639 --> 00:14:42.080 We'll rush the market with our offering, and it was an absolute disaster. 00:14:42.240 --> 00:14:43.360 It was an absolute disaster. 00:14:43.440 --> 00:14:45.440 And then Microsoft decided we've got to jump in. 00:14:45.679 --> 00:14:47.519 Then all the rest started to come to market. 00:14:47.600 --> 00:14:52.320 But what happened was, had they not have launched then, we wouldn't be where we are today. 00:14:52.559 --> 00:14:55.600 It was not yet ready, it wasn't done, it didn't matter. 00:14:55.840 --> 00:14:59.440 It showed enough of the potential of what was there. 00:14:59.679 --> 00:15:15.600 And there's a lesson from this for all of you, which is that in Britain, and I'm a big, big, big, guilty person for this, we obsess, we over-engineer, we over-develop, we spend too long trying to get it perfect. 00:15:15.840 --> 00:15:17.519 The Americans don't do that. 00:15:17.679 --> 00:15:27.279 They slap it out in the market, and they'll go through n iterations of it in a space of three months if they have to, but they'll corner that imagination of the market. 00:15:27.600 --> 00:15:29.039 And that's what happened. 00:15:29.200 --> 00:15:31.679 And so from then, the rest is almost history. 00:15:31.919 --> 00:15:41.440 But the acceleration of AI in the last three months eclipses what happened in every point prior to that three months from when ChatGBT was launched. 00:15:41.600 --> 00:15:46.639 And it's off the charts, and we cannot afford to ignore it. 00:15:46.879 --> 00:15:51.600 The implications of it for universities, for everyone in everyday life is huge. 00:15:51.759 --> 00:15:57.600 We've got to rein in the energy demands and all those things, but it's absolutely massive. 00:15:57.759 --> 00:15:59.120 I hope that helps. 00:15:59.440 --> 00:16:00.080 Absolutely. 00:16:00.240 --> 00:16:09.200 I mean it's really interesting when I hear you speak because it's this intersection of kind of customers, business model, the behaviour, the tech. 00:16:09.600 --> 00:16:18.320 You're able to kind of see across all of those different things, and it's that intersection that I think build really good deep tech businesses. 00:16:18.559 --> 00:16:22.639 Um, I'm obviously interested in your latest deep tech venture. 00:16:22.720 --> 00:16:33.279 Um, as I understand it, you've been ingesting all of this scientific information into Copora, and you know, I think it would be amazing if you could just tell us a little bit about that. 00:16:33.679 --> 00:16:36.399 Yeah, so you know, attention is everything. 00:16:36.559 --> 00:16:37.679 That that was the paper. 00:16:37.840 --> 00:16:43.120 That that that changed how I looked at things, it had a massive influence on what we were doing. 00:16:43.279 --> 00:16:51.360 We were on a parallel path looking to basically understand a massive corpus of information, hence the name corpora. 00:16:52.320 --> 00:16:54.559 Corpora being the plural of corpus. 00:16:55.279 --> 00:17:01.039 And when we set out, we wanted to try and use whatever technologies were around to get a head start. 00:17:01.279 --> 00:17:03.679 So we said, like, we want to do this at scale. 00:17:03.840 --> 00:17:06.000 So we said, what is it at scale? 00:17:06.160 --> 00:17:07.519 What do you mean by that? 00:17:07.839 --> 00:17:11.519 And an avi was, well, someone said, well, is that a few terabytes? 00:17:11.599 --> 00:17:13.279 No, no, no, much bigger than that. 00:17:13.440 --> 00:17:16.240 Well, is it is it maybe a is it a petabyte? 00:17:16.400 --> 00:17:17.680 No, much bigger than that. 00:17:17.839 --> 00:17:22.960 We want to be looking at hundreds of petabytes, maybe even an exabyte one day. 00:17:23.200 --> 00:17:31.200 And we want to be able to therefore take that size of corpus and and understand what's in it, correlate it, graph it. 00:17:31.279 --> 00:17:35.519

So we looked at graph databases, Neo4J, DGraft, you name it. 00:17:35.759 --> 00:17:39.759 We looked at vector databases, we looked at PineCone, all these things coming along. 00:17:40.079 --> 00:17:46.640 None of them would scale to anywhere close past probably a few terabytes, and it's very little further forward today. 00:17:46.720 --> 00:17:51.759 So you have the classic challenge of scaling out is the only way you can go. 00:17:52.000 --> 00:18:01.119 It didn't scale up enough, and if you wanted to scale out from a few terabytes to handle hundreds of petabytes, you're talking about thousands of servers. 00:18:01.359 --> 00:18:03.200 It's an absolute nightmare. 00:18:03.359 --> 00:18:07.440 And if it's a graph, you've got to segment your graph across all those things. 00:18:07.519 --> 00:18:08.400 It's a challenge. 00:18:08.640 --> 00:18:14.079 So we decided, as a bunch of lunatics, we are, that we'd actually set out to look at scale up. 00:18:14.400 --> 00:18:26.559 How can we ingest, understand, fully map, correlate, graph, vectorise, do whatever we need to do with a massive corpus, and we have to do it at speed, otherwise we'll spend 20 years trying to do it. 00:18:26.960 --> 00:18:30.799 So we obsessed over the efficiency of our technology. 00:18:31.039 --> 00:18:36.240 Our entire platform today runs on less than 8 kilowatts of power. 00:18:36.720 --> 00:18:41.279 Entire platform, mapping over 200 petabytes of unstructured data. 00:18:41.839 --> 00:18:45.519 And then the trick was then we wanted to compress that. 00:18:46.160 --> 00:18:50.079 Models compressed data, but they don't expand it again. 00:18:50.480 --> 00:18:55.359 They don't have the ability to give you the attribution of where that information came from. 00:18:55.599 --> 00:18:57.759 We wanted a lossless compression. 00:18:58.240 --> 00:19:10.240 So we've got a compression that gets data down to about one and a half to two percent of its size, but we can fully expand it, knowing exactly where it came from, taking it to the exact article, what it says, where it is. 00:19:10.400 --> 00:19:38.559 And then the next piece was condensing that view of that information into a form that's fully graphed, fully vectorized, all those things that you want to do, and be able to answer questions very precisely by being able to look at all of the totality of information around that topic, corralling it, condensing it, and getting it down to what we call the net unique view of information on that question or topic. 00:19:39.119 --> 00:19:41.359 And we do that in milliseconds. 00:19:41.599 --> 00:19:47.599 Once you've got to that, meanwhile, Chat GPT launches and it threw us into absolute turmoil. 00:19:47.839 --> 00:19:50.960 Oh my goodness, this is gonna change the world. 00:19:51.119 --> 00:19:53.759 This is gonna change how people consume information. 00:19:54.000 --> 00:20:01.519 People are gonna think it's all real and they're gonna get running away with, oh, I've got my nice medical diagnosis from Chat GPT, it looks fantastic. 00:20:01.680 --> 00:20:03.519 It says I've got nothing to worry about. 00:20:03.680 --> 00:20:06.880 Meanwhile, it's all really just a facade of what's there. 00:20:07.200 --> 00:20:27.680 So we said, but if we can take our information, the ability to corral the information of a small topic, a really precise set of questions, and we can pull the net unique view together and then decompose it again to use a small model to be able to represent that, we can make a tiny model look like it's five times the size of Opus 4.7. 00:20:28.240 --> 00:20:30.400 And that's what we built, and that's what it does. 00:20:30.480 --> 00:20:34.799 And we've demonstrated to a few people today that wouldn't stop asking questions. 00:20:34.960 --> 00:20:35.920 Am I right, Helen? 00:20:36.559 --> 00:20:37.039 Absolutely. 00:20:37.119 --> 00:20:38.400 They wouldn't stop get tested. 00:20:38.480 --> 00:20:39.359 Well, how about this then? 00:20:39.519 --> 00:20:40.480 Try that with it. 00:20:40.799 --> 00:21:09.279 And so corporate AI now can take you from a simple challenge statement to produce a really elaborate problem definition to doing a fully ideated solution, to validating that solution, to working out a development roadmap to tell you what the staffing should be, what equipment you need, what the use cases are, what the patent application to protect it all should be, and what the pitch deck is for grant funding or for your VCs. 00:21:10.400 --> 00:21:17.279 And it's basically a complete scientific frontier research platform in a box, one-stop shop. 00:21:17.359 --> 00:21:22.160 Now the entire process, I think less than 30 minutes is a good example of it, Andy. 00:21:22.319 --> 00:21:22.400 Yeah. 00:21:22.480 --> 00:21:22.720 Yeah. 00:21:22.880 --> 00:21:23.119 Yeah. 00:21:23.359 --> 00:21:24.000 Wow. 00:21:24.640 --> 00:21:26.079 That's pretty amazing, right? 00:21:26.240 --> 00:21:35.279 So to bring it to life, are there any challenges or questions that you could share that you've asked it and you've got, you know, things that have excited you? 00:21:35.599 --> 00:21:37.200 Oh, loads, loads. 00:21:37.359 --> 00:21:46.240 So on route to this, funny story, en route to this, uh, there's a lovely guy called Daniel Hayes, and I've got to know him really well. 00:21:46.400 --> 00:21:49.119 And he's now met him on a few occasions as well. 00:21:49.200 --> 00:21:55.759 Uh, and he represented this little group called UCR, Universal Collaboration Research. 00:21:56.000 --> 00:21:59.759 And we weren't ready, the platform wasn't launched, even in its prior incarnation. 00:22:00.319 --> 00:22:02.640 And Daniel said, we'd really love to use your platform. 00:22:03.119 --> 00:22:04.160 We've heard what you're doing. 00:22:04.319 --> 00:22:10.400 We'd like to use it to try and see what we what the patents are in this space, what the intellectual property is. 00:22:10.640 --> 00:22:11.599 And I said, We're not ready. 00:22:11.759 -->

00:22:12.160 We're not ready. 00:22:12.240 --> 00:22:12.880 Go away, we're not ready. 00:22:12.960 --> 00:22:13.359 We're not ready. 00:22:13.440 --> 00:22:14.400 We'll let you know where we are. 00:22:14.559 --> 00:22:18.160 He kept going at it and go, and by the way, that's a great quality. 00:22:18.400 --> 00:22:21.759 For you, the entrepreneurs in the room, do not get put off. 00:22:21.920 --> 00:22:22.799 Learn from kids. 00:22:23.039 --> 00:22:24.640 I want an ice cream daddy. 00:22:24.720 --> 00:22:25.519 You can't have one. 00:22:25.680 --> 00:22:27.119 But I want an ice cream daddy. 00:22:27.200 --> 00:22:30.480 And they'll keep up in the game until you give them an ice cream. 00:22:30.559 --> 00:22:30.880 Okay? 00:22:31.200 --> 00:22:33.839 Now, when you're a grandparent, of course you give it them straight away. 00:22:33.920 --> 00:22:35.039 You don't even wait, do you? 00:22:35.200 --> 00:22:36.480 So that's our job. 00:22:36.640 --> 00:22:37.119 All right. 00:22:37.279 --> 00:22:41.519 So you're in this scenario where you know he kept pushing. 00:22:41.920 --> 00:22:49.119 And he said to us, we want some queries on a thing called, remember it well, multi-layer ceramic bus bars. 00:22:49.279 --> 00:22:56.160 Now we're only a stone throw away from JLR, and some of you in the room probably would actually have an understanding of what that probably means. 00:22:56.319 --> 00:22:56.640 Okay. 00:22:57.119 --> 00:22:58.960 I didn't know much about it at the time. 00:22:59.039 --> 00:23:06.319 We ran some queries, and I had 10 minutes, ran some queries through, sent them to him, and I go, what's that about? 00:23:06.559 --> 00:23:07.759 That's interesting. 00:23:08.079 --> 00:23:12.079 Two minutes later, wow, ten minutes later, can we have a call tonight? 00:23:12.160 --> 00:23:13.359 I think, here we go. 00:23:13.519 --> 00:23:17.759 I told you it wasn't ready, I've done you a favour, and you're just like those guys that worry. 00:23:17.839 --> 00:23:18.799 You just want more. 00:23:18.960 --> 00:23:20.640 You just want to keep going, that's it. 00:23:20.799 --> 00:23:26.079 So we had a conversation, we weren't, it didn't have a license, we had no business model. 00:23:26.240 --> 00:23:28.720 We said, okay, we'll let you use the platform. 00:23:28.960 --> 00:23:42.640 Where they've ended up is in a really interesting, and this is, and by the way, we were so impressed by them, we actually provided some grant funding, we provided some pre-seed funding, they've had less than a million pounds of the funding. 00:23:42.880 --> 00:23:57.119 They now have a working prototype of a device that can take the output of a solar panel, and I'm talking about a current state-of-the-art solar panel, and boost the usable output by 2.4 to 1. 00:23:58.240 --> 00:23:59.599 2.4 to 1. 00:23:59.759 --> 00:24:09.440 It would make solar the cheapest form of energy generation by a country mile, better than fully amortized nuclear, you name it. 00:24:09.920 --> 00:24:14.799 But besides that, they wanted to do that with ceramics, that's where they started. 00:24:15.039 --> 00:24:26.799 They also created a new, more efficient method of creating plasma, sparks that are 7,000 degrees and above, with the amount of energy you get from an electric kettle. 00:24:27.359 --> 00:24:32.160 And with that, they can now produce ceramic materials that can do wonderful things. 00:24:32.319 --> 00:24:41.279 And our problem is they can do so much that our problem is actually, what are we getting all the people to allow us to implement all these things and do these things with it? 00:24:41.519 --> 00:24:45.759 But they've done that, and the platform's been a big part of that journey. 00:24:46.079 --> 00:24:47.039 Does that help? 00:24:47.359 --> 00:24:54.240 Well, yeah, I mean it's not quite cold fusion, but we're getting there, I guess, in terms of um the energy. 00:24:54.400 --> 00:24:58.480 Um does it do, I mean, we talked about the scientific side of things. 00:24:58.559 --> 00:25:04.559 You know, if we were asking it for peacekeeping in the Straits of Hormuz, I mean, can you go in that side? 00:25:04.640 --> 00:25:05.599 Can you go humanitaires? 00:25:05.839 --> 00:25:06.640 Can you go politics? 00:25:06.960 --> 00:25:08.799 Is this guy a plug for Neo? 00:25:09.680 --> 00:25:10.319 Is he? 00:25:11.119 --> 00:25:14.880 Obviously, Neo got to him and said, I didn't quite have enough time to quiz Mel this morning. 00:25:14.960 --> 00:25:16.799 Would you just bring that into the talk? 00:25:16.880 --> 00:25:25.519 You know, so Neo is one of the guys on the peacekeeping force in in Warwick University, and this morning we gave him a license key for what we call curve. 00:25:25.680 --> 00:25:29.119 Curve is the real-time viewer on this data set we have. 00:25:29.279 --> 00:25:35.599 Now, I talked about 200 petabytes, but it's actually looking at changing data sets in the world in real time. 00:25:35.759 --> 00:25:39.359 So every day there's like three or four million new articles for news coming in. 00:25:39.519 --> 00:25:43.119 There's new patents, new SEC filings, you know, all come in real time. 00:25:43.359 --> 00:25:46.799 Curve is the window into that, and Neo wanted to have a license key for that. 00:25:46.880 --> 00:25:47.680 So give him that. 00:25:47.759 --> 00:25:49.680 And he asked exactly that question. 00:25:49.920 --> 00:25:56.000 I said, look, this is a scientific thing for trying to work out problem definitions and coming up with solutions. 00:25:56.240 --> 00:26:00.319 But I tell you what, I'm going to take the risk and try it on these this question. 00:26:00.480 --> 00:26:05.279 And it'll probably blow up in my face, all right, and it probably won't work, but we'll try it. 00:26:05.359 --> 00:26:09.279 I love live demos, they're they're really horrendous, but we do them all, don't we? 00:26:09.359 --> 00:26:10.000 So, you know. 00:26:10.240 --> 00:26:16.160 So, but if you if you pull it off, it's worth a

thousand PowerPoint presentations. 00:26:16.319 --> 00:26:17.759 There could be smoke and mirrors. 00:26:18.079 --> 00:26:26.240 So the challenge was how would we bring about a mediation in a conflict between Ukraine and Russia was the one he gave me. 00:26:26.559 --> 00:26:35.599 So we tried that, and it's come up with a pretty good stab at all the things you'd have to look at and then to come out with how you would actually implement that. 00:26:36.000 --> 00:26:53.359 So it's it's not where I would like it to be, David, on that, but it's a few hours at most to be able to just change the generics to do that, so we can then take on probably the politics if not all the humanities, but certainly the politics side of that. 00:26:53.680 --> 00:26:54.079 Wow. 00:26:54.240 --> 00:27:03.599 I'm um I'm I'm conscious that we haven't got a huge amount of time, so I'm gonna go from that global right down to a sort of very specific question. 00:27:03.759 --> 00:27:07.200 So we've got some amazing deep tech founders in the room. 00:27:07.359 --> 00:27:12.079 I wonder from your own experience what advice you might give to a deep tech founder. 00:27:12.880 --> 00:27:27.359 I think it's interesting because you know, been having been there before and having been there and and failed when I had no money, okay, which is a really difficult place to be because you you don't know where you're going to go with it. 00:27:27.680 --> 00:27:41.759 And having been there where I failed and had enough money to survive that and just regroup and go again, I would say the following that you have to have a pretty good indication, not of feasibility. 00:27:42.000 --> 00:27:46.160 Forget feasibility, go for viability. 00:27:46.559 --> 00:27:48.160 You want viability. 00:27:48.400 --> 00:27:59.200 It's no good coming out with a deep tech innovation that requires 10 times as much energy than another solution but has zero pollution because no one's going to use it. 00:27:59.519 --> 00:28:03.759 So you've got to think: is it actually viable? 00:28:04.079 --> 00:28:09.039 Viable doesn't just mean it works, viable means it's commercially valuable. 00:28:09.279 --> 00:28:13.599 It has a market opportunity that you can identify and go with. 00:28:13.759 --> 00:28:21.359 And then you've got to decide, even if I do this, how does that, how do I get to that go-to-market piece? 00:28:21.599 --> 00:28:24.960 And sometimes it's not something you can do yourself. 00:28:25.200 --> 00:28:30.400 Sometimes you've got to look, well, maybe it's a licensing deal with someone who understands that market. 00:28:30.880 --> 00:28:41.920 Now that might not satisfy your grandiose view of valuation, but it but it's going to give you a fast-track path towards that, and and you probably could get somewhere close as well. 00:28:42.240 --> 00:28:45.440 So think about what it is you're doing. 00:28:45.680 --> 00:28:46.960 Is it feasible? 00:28:47.200 --> 00:28:48.480 Is it viable? 00:28:48.880 --> 00:28:57.519 If

it's viable, what's that bridge between where it is now to go to market and and be wary of that VC trap? 00:28:58.400 --> 00:29:05.359 You can easily spend, if you if you think you've got your idea pretty well sourced, you might even have a bit of a prototype. 00:29:05.680 --> 00:29:15.279 You could waste three, four, five, six months of your time writing documents to try and get a VC to respond to you. 00:29:15.440 --> 00:29:18.960 Even when they do, they'll always say, Yeah, we'll talk to you. 00:29:19.039 --> 00:29:21.920 Let's look at it, we're quite interested in this, and it goes nowhere. 00:29:22.240 --> 00:29:26.319 The emotional roller coaster that brings is huge. 00:29:26.640 --> 00:29:27.440 It's huge. 00:29:27.599 --> 00:29:32.079 And if you talk to four or five of them, and they all say no, you'll doubt yourself. 00:29:32.400 --> 00:29:37.119 If you're confident about what you want to do, have the passion, keep going at it. 00:29:37.200 --> 00:29:38.559 I want an ice cream, Daddy. 00:29:38.720 --> 00:29:40.079 All right, I want an ice cream. 00:29:40.240 --> 00:29:40.960 You can't have one. 00:29:41.039 --> 00:29:44.319 I want an ice cream, and you just keep going at it. 00:29:44.480 --> 00:29:49.200 But don't talk to too many until you figure it out how you do it. 00:29:49.359 --> 00:29:53.759 And don't necessarily treat what they say to you as the things you need to do. 00:29:54.160 --> 00:29:58.160 If it was as simple as that, they'd have backed you and just said, just do these few things. 00:29:58.480 --> 00:30:00.240 So they don't know the market. 00:30:00.559 --> 00:30:03.119 VCs are followers of fashion. 00:30:03.359 --> 00:30:05.839 Look at the AI investments now. 00:30:06.000 --> 00:30:12.960 The people on the fear of missing out that are coming in behind it, they'll probably have no chance of getting many times their money back on it. 00:30:13.519 --> 00:30:15.039 VCs follow fashion. 00:30:15.200 --> 00:30:21.839 If yours, what you've done, is what you've done is really unique, really novel, they won't understand that. 00:30:22.079 --> 00:30:23.440 Don't expect them to. 00:30:23.599 --> 00:30:27.440 So when they ask you, what do you think the total addressable market is of this? 00:30:27.599 --> 00:30:28.559 Walk out. 00:30:29.359 --> 00:30:32.960 If they don't know the market, you're going to aim at well enough to figure out that's huge. 00:30:33.119 --> 00:30:41.599 If you think the guys that did Uber, we've worked out how many taxi journeys there are in the world, and we think actually that the market for this thing is pretty big. 00:30:41.759 --> 00:30:46.240 In fact, in London, it's probably worth about 2,000 a year, maybe 2,000 a month. 00:30:46.480 --> 00:30:47.359 It's bloody huge. 00:30:47.519 --> 00:30:50.160 We're going to disrupt the entire taxi marketplace. 00:30:50.400 --> 00:30:51.680 I don't know how big it is. 00:30:51.920 --> 00:30:53.680 Have you got the ball to the back it? 00:30:53.759 --> 00:30:55.279

That's all I need to know. 00:30:55.519 --> 00:30:56.559 And what did they do? 00:30:56.640 --> 00:31:08.480 They did a smoke and mirror show by arranging for a car to be just around the corner to demonstrate it to, I think it was Bob Cagle from Benchmark at the time, to invest in it to say, look, here's how it works. 00:31:08.640 --> 00:31:10.480 Oh my god, they must be pretty big already. 00:31:10.559 --> 00:31:12.319 The car already arrived. 00:31:12.880 --> 00:31:16.240 But he was sold, invested literally within a week of that. 00:31:16.480 --> 00:31:18.960 He didn't have to go into a big PowerPoint presentation. 00:31:19.119 --> 00:31:22.720 Here's what the cars look like, here's the branding, here's how we're going to market it. 00:31:23.039 --> 00:31:25.279 Going to put an app out there, everyone's going to want to use it. 00:31:25.440 --> 00:31:28.240 And it's disrupted that market massively. 00:31:29.440 --> 00:31:30.079 That's great. 00:31:30.240 --> 00:31:30.720 I love it. 00:31:30.799 --> 00:31:37.039 Um, my last question, because I'm conscious of time, but I promised I'd leave a bit of time for any questions from the audience. 00:31:37.200 --> 00:31:40.079 So start thinking if anybody's got a question. 00:31:40.400 --> 00:31:43.680 My last question is really in this investment space. 00:31:43.839 --> 00:31:49.759 So obviously, bootstrapping versus investment, how do you know when you should go to investor? 00:31:49.920 --> 00:31:50.880 When do you need one? 00:31:51.039 --> 00:31:52.480 What's the trigger for you? 00:31:52.799 --> 00:31:58.319 I I don't think there's a binary view of that or even a litmus test you can apply. 00:31:58.480 --> 00:32:01.920 Uh and I'll take Corporal as a classic example. 00:32:02.160 --> 00:32:05.599 So we saw ChatGPT go to market. 00:32:06.079 --> 00:32:11.440 We saw perplexity, and I don't have a lot of, I don't have a lot of respect for the product. 00:32:11.599 --> 00:32:12.000 They were. 00:32:12.079 --> 00:32:15.440 I think now they're figuring out lots of good things, they're doing some great work. 00:32:15.680 --> 00:32:18.559 But at the time we thought, wow, we need to go in this bandwagon. 00:32:18.640 --> 00:32:22.319 We've got a tool that's better than all these things, we should put it out there in competition. 00:32:22.799 --> 00:32:25.839 We didn't have the resources to market that, to compete with that. 00:32:26.000 --> 00:32:29.359 We were never going to get the the toe hold that was required. 00:32:29.519 --> 00:32:30.880 Well, we'll go and raise some money. 00:32:30.960 --> 00:32:32.640 So we did our business plans around that. 00:32:32.720 --> 00:32:37.279 We went down the VC courtship, and no one wanted to back us. 00:32:37.440 --> 00:32:39.920 And you know, they were absolutely right. 00:32:40.160 --> 00:32:41.359 They shouldn't have backed us. 00:32:41.440 --> 00:32:43.440 It wasn't worth backing on that plan. 00:32:43.759 --> 00:32:50.960 And they did me a big favor because they

forced us to find the route that we could go to market with, and we can now do it without them. 00:32:51.119 --> 00:32:55.359 And now and we've moved from a licensing model, we don't license our technology. 00:32:55.519 --> 00:32:58.480 We might talk about license with Yuli who said that. 00:32:58.559 --> 00:33:00.319 But we're not licensing our technology. 00:33:00.400 --> 00:33:00.720 Why? 00:33:00.880 --> 00:33:06.079 Because the value of what it produces will be democratized by a licensing model. 00:33:06.240 --> 00:33:17.680 And I will not democratize a product that can go from a simple challenge statement to a fully ideated solution with patent application in less than half an hour for 20 bucks a month. 00:33:17.920 --> 00:33:25.039 Because some kid in a bedroom using it will actually put loads of people out of business by just filing patents on the back of that. 00:33:25.279 --> 00:33:26.319 I won't do that. 00:33:26.480 --> 00:33:39.839 So our model is we either allow people to use the technology for a share of the IP, or we invest and we also allow them to use the model the product for a share of the IP, and that's a medium to long-term gain. 00:33:40.000 --> 00:33:40.799 We have no revenue. 00:33:40.880 --> 00:33:42.319 I'm not bothered about that. 00:33:42.480 --> 00:33:45.279 The revenue is just a facade, it's just another measurement. 00:33:45.440 --> 00:33:58.079 But we're building value in allowing companies that can utilize our technology to accelerate their value, to bring their product to market one, two, three years sooner, and for probably a fifth of the cost. 00:33:58.319 --> 00:34:01.039 And in there, we can actually extract value. 00:34:01.359 --> 00:34:04.559 But I mean I'm talking about small bits of value across a broad base. 00:34:04.960 --> 00:34:06.640 And that's the model we're now on. 00:34:07.440 --> 00:34:08.000 Thank you. 00:34:08.719 --> 00:34:10.639 Any questions from the audience? 00:34:11.039 --> 00:34:12.239 I think you first there. 00:34:12.960 --> 00:34:13.760 Thank you very much, Mal. 00:34:14.000 --> 00:34:15.679 It's been inspiring today. 00:34:15.760 --> 00:34:26.239 Uh if you know my question is, um, when and how did you decide to do your philanthropy work, your charity work like in SAR? 00:34:26.400 --> 00:34:26.639 Yeah. 00:34:26.880 --> 00:34:28.159 Where where did it come from? 00:34:28.639 --> 00:34:30.480 I I think just just giving back. 00:34:30.800 --> 00:34:37.119 All right, and if you've ever seen the movie, you know you you'd know exactly what that's that's all about, which is a fabulous movie, by the way. 00:34:37.440 --> 00:34:41.039 So I I I'd made a lot of money over lots of businesses. 00:34:41.280 --> 00:34:44.000 And my attitude was I'd like it to go to some good. 00:34:44.239 --> 00:34:44.559 Right? 00:34:44.639 --> 00:34:52.800 So we created this this trust which has got a very wide remit, medical sporting and research technology, and it has a very wide remit. 00:34:52.880 --> 00:34:54.000 It gets involved in all sorts. 00:34:54.320 --> 00:34:55.920 We sponsor sports people. 00:34:56.239 --> 00:35:01.599 Uh Genevieve Somerville will represent us in the Olympics next year, on the the Winter Olympics. 00:35:01.840 --> 00:35:04.480 Sponsored her since she was about 11 years old. 00:35:04.639 --> 00:35:06.960 Uh we've got tennis players we sponsored. 00:35:07.119 --> 00:35:16.000 We've got someone who was in the Olympics this year, Winter Olympics, who's actually uh basically a solo bobsleigh athlete. 00:35:16.239 --> 00:35:21.519 All right, she lives, she was born in Derby, and uh, you know, and there's no snow there, by the way. 00:35:21.599 --> 00:35:23.519 So you can imagine the work she goes into. 00:35:23.760 --> 00:35:26.639 So we put money in, but they put lots of hard work into it. 00:35:26.719 --> 00:35:38.320 We sponsor medical projects, we sponsored pure technology projects, uh, and we go where we think there's a value to be had, not in a return, but in terms of what's the impact of those investments. 00:35:38.559 --> 00:35:47.119 And and and it's great to do that, and it's it's carried us into all sorts of conversations that I feel honored to be part of because of that. 00:35:50.400 --> 00:35:56.079 So I have a question which is around the conflict between values and interest. 00:35:56.719 --> 00:36:12.320 So if you're a founder and and you found a product where you generally market, why can't we still do it here when you when you just have to try what kind of market in the UK why can't you plan and do it there? 00:36:12.800 --> 00:36:15.679 So is there a conflict between values and what do you think of that? 00:36:16.239 --> 00:36:19.360 I think that uh I'm a big sovereign fan. 00:36:19.760 --> 00:36:20.000 Okay. 00:36:20.480 --> 00:36:36.400 I think that and and you guys are uh on on my journey I see working with universities like this and others, I think, as being a key component of how we actually survive as a nation, how we thrive as a nation. 00:36:36.719 --> 00:36:42.480 And and you know that that means more to me than making 10x what we might make doing that. 00:36:42.639 --> 00:36:45.360 Um so you know, enough is enough. 00:36:45.519 --> 00:36:47.360 One pound more than enough is plenty. 00:36:47.519 --> 00:36:47.760 Okay. 00:36:48.239 --> 00:36:49.679 So you can go to the States. 00:36:49.760 --> 00:36:54.719 That might be the reason you go there, isn't because of the market, it might be because of the backing. 00:36:54.960 --> 00:37:03.760 And I think we've got a lot of work to do in the UK to increase, to accelerate, to make the fit of that backing process better. 00:37:04.000 --> 00:37:06.639 Uh, and I think and we've seen some great initiatives. 00:37:06.800 --> 00:37:10.000 I met with a lady called Josephine Kant three three weeks ago. 00:37:10.159 --> 00:37:13.199 She's one of the founding members of D Sit Sovereign AI. 00:37:13.440 --> 00:37:14.400 Fantastic initiative. 00:37:14.480 --> 00:37:19.840 So it's happening, but the answer is because I want to I want to try and help other things. 00:37:20.320 --> 00:37:21.679 The ground, so values first. 00:37:21.920 --> 00:37:22.000 Yeah. 00:37:24.639 --> 00:37:26.400 I mean, it's a great place to wrap it up. 00:37:26.480 --> 00:37:29.440 I have to say, it's fantastic to hear Mel speak, isn't it? 00:37:29.519 --> 00:37:31.760 What an amazing, inspirational person to have in. 00:37:31.920 --> 00:37:34.320 So I think he deserves a round of applause.