00:00 can outsource your thinking to AI, but
00:02
00:02 you can’t outsource your understanding.
00:04
00:04 I’m all about companies, even if they’re
00:06
00:06 called tech companies, they’re all about
00:07
00:07 people. So, my first investors that I
00:09
00:09 presented to, they said we didn’t win
00:11
00:11 and there was more than 100
00:13
00:13 that I presented to and then they all
00:14
00:14 said no.
00:15
00:15 The main reasons were they said, “Look,
00:17
00:17 we need to believe in two miracles to
00:18
00:18 happen for us to invest in you.” It’s a
00:20
00:20 founder skill that AI cannot replace.
00:25
00:25 Hello and welcome to Tomorrow the Future
00:28
00:28 Forum podcast. I’m Oliver Barrett and
00:31
00:31 here at Future Forum we’re all about
00:33
00:33 bringing people together from very
00:34
00:34 different industries to shape the future
00:36
00:36 together.
00:38
00:38 Today’s guest is one of the UK’s
00:41
00:41 defining success stories in
00:43
00:43 entrepreneurship. He co-founded Onfido,
00:46
00:46 building it into one of Europe’s
00:48
00:48 standout technology success stories. So,
00:51
00:51 today’s about identity in more ways than
00:54
00:54 one. This is someone whose journey spans
00:56
00:56 multiple countries and multiple
00:58
00:58 ventures. It’s about how ambitious
01:01
01:01 people keep moving. It’s about what the
01:03
01:03 future of building organizations looks
01:06
01:06 like. Husayn Kassai, welcome to
01:08
01:08 Tomorrow. It’s great to be here. Now,
01:10
01:10 Husayn, so many things to discuss today,
01:13
01:13 but first,
01:14
01:14 give us a glimpse of the future. It’s
01:16
01:16 something you must think about a lot.
01:18
01:18 When you look out there, what do you
01:20
01:20 see? What’s coming?
01:22
01:22 In my mind, having built an AI in one
01:25
01:25 way or another since 2010, it’s
01:27
01:27 naturally fast forward to 16 years from
01:29
01:29 now if I rewind 16 years.
01:31
01:31 I 16 years from now, I have a strong
01:35
01:35 sense that we are constantly going to be
01:38
01:38 as knowledge workers, especially in a
01:39
01:39 high-judgment environment, very
01:42
01:42 regularly interacting with AI in one
01:44
01:44 form or another.
01:45
01:45 And we will in part enjoy that and in
01:49
01:49 part still be times where we get
01:50
01:50 frustrated. And we will yearn for uh you
01:54
01:54 know how you have technology detox?
01:57
01:57 There’ll be camps and experiences and
01:59
01:59 and venues that are just like full-on
02:01
02:01 technology and AI
02:03
02:03 switching off. So, you get to go back to
02:05
02:05 a space where
02:07
02:07 kind of like in in the states in New
02:08
02:08 York, you can go to a jazz bar that has
02:09
02:09 like a 1920s vibe. I think that kind of
02:12
02:12 thing is going to come back. So, it’s
02:13
02:13 going to
02:13
02:13 being in non-AI environment in 16 years
02:16
02:16 from now is going to be quite the thing.
02:18
02:18 Amazing. So, let’s unpick some of that.
02:21
02:21 But speaking of unpicking,
02:23
02:23 to what extent
02:25
02:25 is your entrepreneurial journey
02:28
02:28 intertwined with your technology
02:30
02:30 journey?
02:32
02:32 I I’m interested in solving problems and
02:36
02:36 how essentially
02:38
02:38 my view is like prob- there are certain
02:41
02:41 problems that that have always existed.
02:42
02:42 It’s almost always can something be done
02:44
02:44 better, faster, at a lower cost.
02:46
02:46 And then there are different
02:47
02:47 technological building blocks that come
02:49
02:49 that suddenly make that possible. So, we
02:51
02:51 always would prefer to have booked a
02:53
02:53 taxi faster and then for us to
02:55
02:55 track it. But then a GPS and internet
02:57
02:57 and smartphones made like an Uber
02:59
02:59 happen. Same with an experience on
03:01
03:01 Airbnb. Same So, a lot of these these
03:02
03:02 tools
03:03
03:03 So, applications are there as a result
03:05
03:05 of the technological building blocks
03:07
03:07 finally coming together in [snorts]
03:09
03:09 order to unlock something. And it
03:10
03:10 doesn’t have to be technology. Sometimes
03:12
03:12 it could be regulation. So, the rise of
03:13
03:13 fintech in large part was a combination
03:16
03:16 of smartphones and a regulation such as
03:18
03:18 the virtual licenses for banking and and
03:20
03:20 beyond. So, in that sense, I from sort
03:23
03:23 of a young age I was I was in- my first
03:25
03:25 company was a business was like on eBay.
03:27
03:27 That was kind of new back in this is
03:29
03:29 2001, 2002. So, then just give us a
03:32
03:32 pinpoint on your age cuz that is young.
03:34
03:34 12-ish. 12, yeah. 12, 13. And my first
03:37
03:37 company was on eBay
03:39
03:39 sort of site or eBay service. What do
03:41
03:41 you call it? eBay marketplace account.
03:44
03:44 Along with a PayPal account and I kind
03:47
03:47 of got going. And it was e- So,
03:49
03:49 e-commerce was like the first first
03:52
03:52 first part of my journey.
03:53
03:53 In that university, I tried a few
03:54
03:54 different things. Obviously, identity
03:56
03:56 was was always like stuck with me as a
03:57
03:57 problem that certainly needs to be
03:59
03:59 solved. And I got exposure to machine
04:02
04:02 learning computer vision then.
04:04
04:04 Uh alongside uh my co-founder, my
04:06
04:06 technical co-founder, his university
04:08
04:08 thesis was using computer vision and
04:10
04:10 machine learning to spot wildlife in a
04:12
04:12 series of photos of the jungle.
04:13
04:13 Back in 2010 in Oxford, there wasn’t
04:15
04:15 very much happening in machine learning
04:17
04:17 computer vision at all. Uh the building
04:19
04:19 next door uh had DeepMind in
04:22
04:22 and uh well, it was tiny team. No No one
04:24
04:24 really knew about them until 2011 12 and
04:26
04:26 and beyond. But um in my mind, it was
04:31
04:31 more than more interestingly, it was
04:32
04:32 like this is very clearly the future.
04:34
04:34 Like it was very basic. It’s like got a
04:36
04:36 long way to go.
04:38
04:38 But I was so keen to solve the identity
04:40
04:40 problem that using the latest
04:42
04:42 technology, even though it was like very
04:44
04:44 very uh completely different to what
04:45
04:45 what you what you have today, uh that
04:47
04:47 kind of clicked for me. So, my [snorts]
04:49
04:49 interest was solving problems and it’s
04:51
04:51 always what’s the latest technology that
04:52
04:52 can help best solve that problem. Well,
04:54
04:54 there’s a brief conundrum there. What’s
04:56
04:56 your advice because today’s 12-year-olds
04:59
04:59 are surrounded by even more technology
05:02
05:02 and some of the time we say let’s keep
05:04
05:04 them as far away from it as possible.
05:06
05:06 Uh your story gives us slightly
05:08
05:08 different clues or maybe not.
05:11
05:11 >> I was fortunate enough to have the
05:13
05:13 freedom
05:14
05:14 to be able to pick and choose and
05:16
05:16 experiment uh quite a bit.
05:18
05:18 So, I would say uh I’m a parent myself.
05:20
05:20 I’ve got two very young kids. And I’m
05:23
05:23 more in the camp of
05:25
05:25 taking away technology completely.
05:27
05:27 Uh or as as much as one can.
05:30
05:30 It for the what you really want for
05:32
05:32 under-18s is for them to develop a fully
05:36
05:36 cognitive functioning uh well-rounded
05:39
05:39 experience.
05:40
05:40 And what they will have after they’re 18
05:43
05:43 is a
05:43
05:43 good 100 years of exposure to
05:45
05:45 technology. I I we will comfortably live
05:47
05:47 into well into our hundreds, I believe,
05:49
05:49 with advances in AI, especially on
05:51
05:51 healthcare, longevity, and beyond,
05:52
05:52 subject to us getting good nutrition
05:54
05:54 and, you know, all all the other things.
05:56
05:56 So, as such, there’s no rush, and I
05:58
05:58 worry that some parents now,
06:00
06:00 understandably, have anxiety around the
06:02
06:02 future of work, and they think the
06:03
06:03 logical thing might be to them, “Let’s
06:05
06:05 start our children on an earlier path to
06:08
06:08 learning how to code, learning how to do
06:10
06:10 a Let’s let’s fill their time with like
06:13
06:13 100 different activities and and sort of
06:15
06:15 language lessons and beyond. Where I
06:17
06:17 [snorts] say that’s kind of not needed.
06:19
06:19 The the research, the studies, the
06:20
06:20 experience, and beyond suggests that you
06:22
06:22 just need them to be kids. And what does
06:25
06:25 that mean? Playtime. And, frankly,
06:27
06:27 non-tech things where where relevant and
06:28
06:28 where possible.
06:30
06:30 Um after-school clubs.
06:32
06:32 Anything physical uh
06:34
06:34 that they can sort of um sort of use
06:36
06:36 their physical muscles and and
06:38
06:38 interacting with other sort of children
06:39
06:39 of their age and beyond. All that to me
06:41
06:41 is the crucial parts.
06:42
06:42 >> Well, let’s crystallize that, then. What
06:44
06:44 do you make of the thought of today, a
06:46
06:46 12-year-old embracing technology to
06:49
06:49 start a business? That was you?
06:51
06:51 >> [snorts]
06:52
06:52 >> There’s nothing wrong with that. I What
06:54
06:54 So, if you want to experiment and and
06:56
06:56 try um I still don’t know if it’s legal.
06:60
06:60 Uh I think you might be able to do a
07:01
07:01 paper round at 14. I’m not not not too
07:03
07:03 sure, but what I used my uh older
07:05
07:05 sister’s uh details to be able to start
07:08
07:08 my eBay business and PayPal account, but
07:10
07:10 there’s actually there’s some good in uh
07:13
07:13 some of these games and some of the
07:14
07:14 online activity. You kind of develop
07:16
07:16 certain skills, and that’s fine. It’s
07:18
07:18 everything is There’s a balance to be
07:19
07:19 had. I believe in China they cap it at
07:22
07:22 an hour of of of social media and like
07:24
07:24 online or or smart screen.
07:26
07:26 And that’s [snorts] a pretty
07:27
07:27 heavy-handed approach, but I would say
07:30
07:30 now that we know the studies and the
07:31
07:31 research that suggests all the negative
07:33
07:33 consequences,
07:35
07:35 you could [snorts] kind of understand
07:36
07:36 why a heavy-handed approach is needed,
07:38
07:38 especially on social media. So so, using
07:41
07:41 tools to develop and and have a startup
07:43
07:43 is perfectly good as long as it’s
07:44
07:44 balanced, but but having kids in front
07:46
07:46 of social media or games or like frankly
07:48
07:48 just like the TV or any sort of smart
07:50
07:50 screen for hours on end is probably the
07:52
07:52 by far the worst thing you could do for
07:53
07:53 your kids. Right. So, let’s go through
07:55
07:55 this whistle-stop journey. You’re
07:57
07:57 co-founding a company. Your neighbors
07:58
07:58 are
07:59
07:59 DeepMind.
08:01
08:01 This is Oxford University where you met
08:03
08:03 your co-founders. You scale it into 500
08:05
08:05 people plus. You exit it for hundreds of
08:08
08:08 millions of dollars. How does that
08:10
08:10 happen? Because unlocking the
08:13
08:13 investment, the clients, these are
08:16
08:16 things that some people would have said,
08:18
08:18 “Husayn,
08:19
08:19 you’re going to be
08:21
08:21 defeated every stage by billion-dollar
08:23
08:23 companies along the way, and yet you
08:25
08:25 won.”
08:26
08:26 Yep. We There was a technological
08:29
08:29 paradigm shift and a behavioral paradigm
08:31
08:31 shift that we recognized and others and
08:34
08:34 incumbents, the existing sort of
08:36
08:36 companies that could have done it,
08:37
08:37 recognized far too late. And others
08:41
08:41 relative to us either didn’t realize it
08:43
08:43 or if they did realize it, we worked
08:45
08:45 hard to outcompete them.
08:47
08:47 And [snorts] that is the essence of
08:48
08:48 strategy in some ways. Also, the
08:50
08:50 innovator’s dilemma as far as these
08:52
08:52 internal sort of existing incumbents
08:54
08:54 couldn’t do it. What happened? So,
08:56
08:56 registered the company 2010. I was still
08:58
08:58 a student. I only started working on it
08:60
08:60 full-time in the summer of 2012 when I
09:02
09:02 graduated. And then a year later, my two
09:04
09:04 co-founders were able to join. One
09:06
09:06 graduated, the other one We had
09:07
09:07 customers and investors so he could
09:09
09:09 essentially quit his job and and still
09:11
09:11 continue to pay the mortgage.
09:13
09:13 >> [snorts]
09:13
09:13 >> So, we really started going in 2013. And
09:15
09:15 then the timing was very fortunate.
09:17
09:17 You’ll remember really well. 2014
09:20
09:20 was when the trust marketplaces came
09:21
09:21 about. And on-demand nannies, tutors,
09:24
09:24 doctors, cleaners, and so forth. And you
09:25
09:25 really needed to verify their identity.
09:27
09:27 >> [snorts]
09:28
09:28 >> So, we built our technology which was
09:29
09:29 the right time at the right place.
09:31
09:31 And the trust marketplaces were not seen
09:33
09:33 as a serious industry to begin with. But
09:36
09:36 we kind of could see the future in that
09:37
09:37 and built it. And then that helped the
09:39
09:39 technology get strong enough for us to
09:40
09:40 then in 2015 double down on a fintech
09:43
09:43 space, which again is very online fraud
09:45
09:45 was key. So, at every sort of step of
09:47
09:47 the journey
09:49
09:49 we were fortunate in our timing and in
09:50
09:50 some ways both the technology and the
09:52
09:52 go-to-market approach kind of clicked.
09:54
09:54 And just to pause there, these are
09:56
09:56 platforms where you have to effectively
09:58
09:58 prove that you are who you say you are.
10:00
10:00 And this is a combination of taking
10:02
10:02 photos, showing passport documentation,
10:04
10:04 knitting these pieces of information
10:06
10:06 together. Correct. What existed prior to
10:09
10:09 us was primarily the identity space was
10:11
10:11 very small in the
10:14
10:14 use case of regulation. So, if you
10:16
10:16 needed a financial sort of service that
10:18
10:18 used a credit bureau
10:20
10:20 um where they verify that your name,
10:22
10:22 date of birth, address kind of existed.
10:24
10:24 And that doesn’t stop fraud. It really
10:26
10:26 didn’t make much sense to me.
10:28
10:28 Uh so, what we knew that this new world
10:30
10:30 with a smartphone is going to be require
10:32
10:32 trust to be built online and in digital
10:34
10:34 space. And [clears throat] as an offline
10:37
10:37 person, I need to be able to verify that
10:39
10:39 I am who I claim to be online. And the
10:41
10:41 approach to that is a photo of your
10:42
10:42 passport, a photo of your face, which we
10:44
10:44 all do now on a regular basis. By the
10:46
10:46 time it was novel.
10:48
10:48 And that the there were two things
10:52
10:52 miracles that we had to prove. So, my
10:54
10:54 first investors that I presented to they
10:56
10:56 said we wouldn’t and there was more than
10:57
10:57 100
10:59
10:59 that I presented to and then they all
10:60
10:60 said no.
11:01
11:01 And the main reasons were they said,
11:02
11:02 “Look, we need to believe in two
11:04
11:04 miracles to happen for us to invest in
11:05
11:05 you. The first is a technological one.
11:08
11:08 You’re telling us a photo of a passport
11:10
11:10 and a face essentially
11:12
11:12 technology can do a better job of
11:13
11:13 detecting if this passport is fake or
11:14
11:14 not. Like, we can’t see this.” This is
11:17
11:17 You got to remember this back in 2012.
11:19
11:19 Um the building blocks came together
11:21
11:21 then, by the way. Camera phone
11:23
11:23 and quality of the new smartphones got
11:25
11:25 better and tech connectivity went 3G and
11:27
11:27 then 4G
11:28
11:28 uh internet connectivity, but also cloud
11:30
11:30 service AWS and others meant you could
11:32
11:32 run
11:33
11:33 narrow AI or machine learning computer
11:34
11:34 vision models in a more affordable way.
11:37
11:37 But back to the technological
11:38
11:38 bottleneck, they they couldn’t when we
11:41
11:41 said in theory it’s possible, but we had
11:43
11:43 to prove it and then we couldn’t until
11:44
11:44 they gave us the money and then so
11:46
11:46 forth. But the second miracle that I
11:47
11:47 believe in was a behavioral one.
11:50
11:50 And how, you know, we’re all used to
11:51
11:51 going inside a bank branch. We do it
11:54
11:54 once every 5 years. We, you know, faff
11:56
11:56 around with paperwork. We finally get
11:57
11:57 the job done and then it’s like baked
11:58
11:58 into what we have to do. You’re
11:60
11:60 expecting us for people to sit at home
12:01
12:01 and and get a phone out and download an
12:03
12:03 app to do their banking? Like that’s
12:04
12:04 just really bizarre. And you expect them
12:06
12:06 to take a photo and even taking a
12:08
12:08 selfie? I remember my early investors
12:09
12:09 just wouldn’t even buy into the selfie
12:11
12:11 thing. Now, they are though everyone
12:13
12:13 like doing selfies all the time.
12:14
12:14 >> Yeah, you really non-investors. Yeah, I
12:15
12:15 should say my investors I was I was
12:17
12:17 hoping to get.
12:19
12:19 And therefore,
12:20
12:20 they were kind of very fair. And in
12:22
12:22 terms of 2012
12:23
12:23 mindset, you should they should you
12:25
12:25 should question is this technology be
12:27
12:27 possible.
12:28
12:28 And equally, it will be a it be a
12:30
12:30 behavioral change because arguably
12:32
12:32 behavior change is very difficult. But
12:34
12:34 we felt this is such a key problem and
12:36
12:36 that the it’s really going to be a
12:38
12:38 bottleneck to this amazing series of
12:40
12:40 future services that we ought to have
12:41
12:41 digitally and therefore let’s let’s try
12:44
12:44 talking to more investors see if someone
12:45
12:45 eventually invests. So, which they did.
12:48
12:48 You scale the company, you exit that
12:50
12:50 company. So, let’s step back. If you
12:52
12:52 think about how you now build companies
12:55
12:55 and indeed how life is changing for
12:57
12:57 founders, what is easier than it has
13:01
13:01 ever been and what has become more
13:03
13:03 difficult?
13:05
13:05 Easier is probably
13:08
13:08 there’s
13:09
13:09 much more investor funds out there
13:12
13:12 than there was back in 2012.
13:15
13:15 I could count the number of like VCs,
13:18
13:18 venture capitalists that would like look
13:20
13:20 at companies on basically like a one
13:21
13:21 hand.
13:22
13:22 They were more, but one that would like
13:24
13:24 literally take a a with you and and and
13:26
13:26 like consider we’re doing. Now, there
13:28
13:28 you know, very many.
13:30
13:30 I’ve lost count. I think there I believe
13:31
13:31 there are 3,000 VCs like in the world.
13:34
13:34 And angel investors again it’s become a
13:35
13:35 whole industry. So, but that’s part of
13:38
13:38 it. I think more importantly though for
13:40
13:40 founders it’s actually access to
13:42
13:42 incredible talent because there have
13:44
13:44 been a wave of a series like different
13:46
13:46 startups have scaled and people inside
13:48
13:48 these startups have like learned
13:50
13:50 the playbooks.
13:51
13:51 And now if you want to specific skill
13:53
13:53 set, you know, like a UX with a security
13:56
13:56 background type of profile, you may have
13:59
13:59 a few dozen options and maybe back in
14:01
14:01 2013-14 you would have had maybe five.
14:04
14:04 So, that that by far is the best thing
14:05
14:05 that access to talent. Um,
14:08
14:08 what is harder than it was before,
14:10
14:10 [clears throat]
14:10
14:10 you have a significant greater intensity
14:13
14:13 of competition. So, startups were not a
14:16
14:16 main thing to do in 2012. There were
14:18
14:18 very relatively few startups and there
14:21
14:21 were far fewer AI startups or machine
14:23
14:23 learning I can I can count the the main
14:25
14:25 ones in early 2000 2010 like
14:29
14:29 2012-2013-2014
14:30
14:30 like Darktrace,
14:32
14:32 um, Tractable came a bit later. There
14:34
14:34 was obviously DeepMind.
14:36
14:36 And and they became like really known in
14:38
14:38 a mainstream way like 2014 on anyway.
14:41
14:41 And it was like us as Onfido maybe a
14:43
14:43 couple more. Um, but regardless of that
14:46
14:46 we were again supported by amazing
14:47
14:47 machine learning AI talent from
14:48
14:48 university PhDs and others like that.
14:50
14:50 So, so talent is now easier. What is
14:53
14:53 harder is this intensity of competition.
14:55
14:55 Uh, given that there’s so many more and
14:58
14:58 in an early stage company if you’re
14:59
14:59 prior to product market fit, you could
15:02
15:02 spend 18 months or two years getting
15:04
15:04 there a decade ago because uh
15:07
15:07 early customers had no choice frankly.
15:09
15:09 They’d have to just be patient and wait.
15:10
15:10 Whereas now if you don’t do a good
15:12
15:12 enough job in a year,
15:14
15:14 uh, you might miss the boat and there’ll
15:15
15:15 be 10 others that will be able to which
15:16
15:16 in many ways is a great thing. But as
15:19
15:19 from a founder perspective, the
15:20
15:20 intensity of competition is much greater
15:22
15:22 which which means you have to work even
15:23
15:23 harder. Right. And on that first point
15:26
15:26 about talent,
15:27
15:27 you’re building a new company, Allo, and
15:29
15:29 we can come on to
15:30
15:30 that.
15:33
15:33 You’re also surrounded by a whole suite
15:35
15:35 of AI tools which can perform certain
15:37
15:37 tasks. So, how do you navigate the
15:41
15:41 decision when you’re thinking, “Do I
15:43
15:43 hire a person, or can the suite of
15:45
15:45 things I’m looking at to be done be done
15:47
15:47 by technology, to be done by an AI?”
15:51
15:51 In our in the in the startup building
15:54
15:54 space, you’re primarily dealing with
15:56
15:56 knowledge workers,
15:58
15:58 and where high judgment is required. So,
16:00
16:00 you certainly need tools,
16:04
16:04 and essentially that augments the the
16:07
16:07 people that you hire. So, in essence, my
16:10
16:10 whole philosophy on AI is AI is a
16:13
16:13 capability that you can’t buy, but a
16:15
16:15 capability you have to build.
16:17
16:17 And that ought to be centered around key
16:20
16:20 team members that are then augmented by
16:22
16:22 AI, as opposed to centered around an AI
16:25
16:25 tool that that essentially
16:27
16:27 you have junior team members working
16:28
16:28 around. So, what that means is we’re now
16:30
16:30 a team of 14, and there
16:33
16:33 most are mid-level skilled or or more
16:36
16:36 senior than usual, but they’re able to
16:38
16:38 produce two or three times what would
16:39
16:39 have been needed, say, a decade ago.
16:42
16:42 Because they’re now augmented by all
16:43
16:43 these tools. So, said another way, my
16:45
16:45 team of 14 now are probably doing what a
16:47
16:47 team of 50 people would have done
16:49
16:49 a decade ago. So, fewer but more skilled
16:52
16:52 team members because they’re able to do
16:54
16:54 a lot more with these AI tools. We have
16:56
16:56 to be very good at recognizing what do
16:58
16:58 you want the AI tools to deliver. And AI
17:01
17:01 tools are wonderful at pattern
17:03
17:03 recognition and scaling that sort of
17:06
17:06 work, but
17:07
17:07 frankly don’t do well in judgments at
17:09
17:09 all.
17:10
17:10 And Professor Vervaeke in the in Canada
17:13
17:13 talks a bit about different types of
17:14
17:14 intelligence. There’s propositional
17:16
17:16 intelligence, which is facts and
17:18
17:18 essentially knowing all of Wikipedia and
17:19
17:19 beyond. AI is excellent at that.
17:22
17:22 Then you have procedural intelligence,
17:24
17:24 the steps needed to do something. AI is
17:25
17:25 good and getting better, but has some
17:27
17:27 gaps.
17:28
17:28 And then you have perspectival
17:29
17:29 intelligence. And that is from my
17:31
17:31 perspective in a sitting in a room,
17:34
17:34 reading the room, how how are things
17:36
17:36 done and and and so forth. That AI can’t
17:39
17:39 do well at all and will, in my view,
17:40
17:40 continue to struggle.
17:42
17:42 And that’s why last week I heard this
17:44
17:44 this phrase which really resonated
17:45
17:45 around how you can outsource your
17:47
17:47 thinking to AI, but you can’t outsource
17:49
17:49 your understanding. And understanding is
17:51
17:51 all about let’s how does something work
17:53
17:53 behind it? What are the So even the
17:55
17:55 latest models right now, they can’t look
17:57
17:57 at a clock and tell you the time in
17:58
17:58 effect like a
17:59
17:59 traditional analog clock.
18:01
18:01 And so we must not underestimate the
18:03
18:03 human cognitive power.
18:05
18:05 And the best AI companies, or excuse me,
18:07
18:07 best companies across the board will get
18:08
18:08 the right balance. Let’s build AI
18:10
18:10 capabilities in-house and outsource the
18:12
18:12 pattern recognition components to to
18:15
18:15 sort of AI tools.
18:16
18:16 So how has that evolution in terms of
18:19
18:19 what you’re looking for
18:21
18:21 affected how you find and filter the
18:25
18:25 talent that’s applying or the talent
18:26
18:26 that you go out and find? Because this
18:28
18:28 is a different role that they’re now
18:30
18:30 performing. Yeah. So in a few ways. The
18:33
18:33 The main is
18:35
18:35 going back to my point around education
18:37
18:37 for under-18s and and being
18:38
18:38 well-rounded.
18:39
18:39 If you look at the bits that AI can’t do
18:41
18:41 well, there’s um Dr. Pink’s book around
18:45
18:45 left and right hemisphere of the brain
18:47
18:47 and essentially how right hemisphere
18:49
18:49 brain thinkers will will rule the
18:51
18:51 future.
18:52
18:52 And in short, the left hemisphere is
18:53
18:53 very strong on analytics and frankly
18:56
18:56 pattern recognition. I’m simplifying,
18:59
18:59 but that’s the a good way one could
19:01
19:01 think about it. And the right hemisphere
19:03
19:03 is a lot of things that we would
19:04
19:04 associate with with what makes us human.
19:06
19:06 That that is connecting the dots.
19:08
19:08 And that is a bit more on the EQ,
19:11
19:11 emotional intelligence,
19:12
19:12 understanding, reading room,
19:14
19:14 relationships and and beyond.
19:16
19:16 And when you look at hiring someone, you
19:18
19:18 want them to be able to connect with us.
19:20
19:20 You want them to have be able to have
19:21
19:21 and develop the judgment. You want them
19:22
19:22 to be able to build relationships. So,
19:24
19:24 these things are not things you can test
19:26
19:26 for easily. So, the standard test for
19:28
19:28 most of the plan to banks and consulting
19:30
19:30 firms is is frankly like a math and an
19:31
19:31 IQ based test, which is completely the
19:33
19:33 opposite of what they Not the opposite,
19:35
19:35 but completely misses the big half of
19:37
19:37 the picture.
19:38
19:38 >> [snorts]
19:38
19:38 >> And as such, when we hire, it’s normally
19:40
19:40 good especially if they’re junior, they
19:42
19:42 just come and spend a few days with us
19:44
19:44 in the office so that they You have to
19:45
19:45 It’s an experience that you sort of feel
19:48
19:48 after And you what you’re looking for is
19:50
19:50 potential. So, what I answered is a
19:52
19:52 little bit about like like cognitive
19:54
19:54 capabilities. But
19:56
19:56 how we put that in a structure and how
19:58
19:58 does the team what did actually look
19:59
19:59 for? There’s three parts.
20:02
20:02 Humble, hungry, and high high potential.
20:04
20:04 So, humble in the sense that they are is
20:07
20:07 frankly like a proxy for EQ.
20:09
20:09 And that they don’t put their
20:11
20:11 selves first, they put the mission, the
20:13
20:13 company first. They’re they’re good team
20:14
20:14 player. You can be a colleague of
20:15
20:15 someone who’s humble [clears throat] as
20:17
20:17 opposed to someone who’s perhaps very
20:18
20:18 arrogant.
20:19
20:19 Um hungry in the sense that they have a
20:22
20:22 high bar of excellence, they’re going to
20:23
20:23 continuously work to improve. They are
20:25
20:25 not here to take a box and clock out,
20:26
20:26 they’re here to like change the world.
20:28
20:28 And then the third is high potential is
20:30
20:30 probably the most misunderstood. High
20:32
20:32 potential by no means means someone with
20:34
20:34 like A stars or like pedigreed
20:36
20:36 universities. It means someone with the
20:38
20:38 capacity to learn.
20:39
20:39 And often this is the bit that’s missed
20:41
20:41 the most. You kind of find the right
20:42
20:42 profile of someone which may have not
20:44
20:44 gone to university even. Uh but shows
20:47
20:47 great ability and enthusiasm and the
20:48
20:48 ability to learn. So, I don’t know your
20:51
20:51 question is around in the age of AI, you
20:53
20:53 kind of want those who are the fast
20:54
20:54 learners with all these other things.
20:56
20:56 The last thing to say is
20:57
20:57 um we just hired our first 21-year-old
21:01
21:01 uh 3 weeks ago.
21:02
21:02 And [snorts] it was part like telling
21:03
21:03 the team, guys like we must must start
21:06
21:06 bringing in junior people in a large
21:07
21:07 part because I feel this is the thing
21:09
21:09 that’s missed amongst large companies
21:11
21:11 the most.
21:13
21:13 Our company I allow we just focus on
21:15
21:15 companies of 2,500 employees and fewer.
21:18
21:18 But my advice to companies that are
21:19
21:19 larger, and the larger you are the
21:21
21:21 bigger the problem is, in that
21:24
21:24 it’s very very difficult to build AI
21:26
21:26 capabilities
21:27
21:27 in a structured way. And AIs is going to
21:31
21:31 change the way you work and that means
21:32
21:32 that just gets complex when you want to
21:35
21:35 manage a hot large organization. The
21:37
21:37 number one biggest thing you can do to
21:38
21:38 change that over the next decade is to
21:40
21:40 make sure your new joiners a greater
21:42
21:42 proportion are frankly under 25. Yeah.
21:45
21:45 Cuz so if you come under 25, you don’t
21:46
21:46 need to teach them AI.
21:47
21:47 >> [laughter]
21:48
21:48 >> They come in like embedded with AI. And
21:50
21:50 if you sit them behind a desktop and you
21:52
21:52 try to show them a process, take this
21:53
21:53 folder to this file and then type this,
21:55
21:55 they’ll be like, “What on earth are we
21:56
21:56 doing here? Did you all know there is
21:58
21:58 AI?” Yeah, exactly.
21:59
21:59 >> a critical mass of them in, they’ll
22:01
22:01 force over a decade AI adoption across
22:03
22:03 the company. Yeah. I mean, that is a an
22:06
22:06 optimistic vision,
22:08
22:08 isn’t it, of the 20-somethings that
22:10
22:10 you’re taking in? And it does depend on
22:12
22:12 them having been immersed in those
22:15
22:15 tools. Unless you think that’s
22:16
22:16 inevitable.
22:17
22:17 >> you’re you’re you’re completely correct.
22:18
22:18 And but but
22:20
22:20 and in large way that’s that’s Allus
22:21
22:21 philosophy in that you have to have the
22:23
22:23 technological tools matched to the
22:25
22:25 trained human expertise. And you should
22:27
22:27 have those human experts trained inside
22:29
22:29 your organization, not outsourced. And
22:31
22:31 and like you have to develop that
22:32
22:32 yourself. Because if you bring them in
22:35
22:35 and you don’t give them the tools, what
22:37
22:37 will happen is they’ll just use their
22:38
22:38 smartphone. They’ll a third of employees
22:40
22:40 are using shadow AI at their companies.
22:43
22:43 So just say what that means, shadow AI.
22:44
22:44 >> Shadow AI is basically means
22:45
22:45 I need to do a task and I’m going to
22:48
22:48 look around my shoulders to make sure no
22:49
22:49 one’s looking. I’m going to take a photo
22:50
22:50 of the screen of what my task is. I’m
22:52
22:52 going to get like Gemini or something to
22:54
22:54 transcribe it. Then I’m going to get
22:55
22:55 Claude or something like to solve it.
22:57
22:57 Then I’m going to type the answer.
22:59
22:59 And that is I’m I’m doing something I
23:00
23:00 should not be doing. I’m using shadow AI
23:02
23:02 cuz that tool is not sanctioned by the
23:04
23:04 company.
23:04
23:04 >> And so, Shadoware is a big problem
23:06
23:06 because uh unless you’re giving the best
23:08
23:08 technology, they’re going to at home
23:11
23:11 This isn’t happened with other
23:12
23:12 technologies before, but people are
23:14
23:14 getting trained at home on evenings and
23:16
23:16 weekends all the time on AI. Cuz when
23:18
23:18 they want to plan a birthday or or buy
23:20
23:20 something or go on travels, like they’re
23:21
23:21 using the best AI tools cuz these AI
23:23
23:23 tools are available for free. And then
23:25
23:25 they come to work and they’re like,
23:26
23:26 “Okay, we’re using old technologies.”
23:27
23:27 You and I would be frustrated, right? If
23:29
23:29 we were suddenly told to go use a fax
23:30
23:30 machine instead of email someone, we’d
23:32
23:32 be like, “No, I’m going to email them.”
23:33
23:33 >> Mhm. Uh
23:34
23:34 hence
23:35
23:35 um that’s but the equally important is
23:38
23:38 you need your AI technology to also have
23:42
23:42 context around the company. Needs to
23:43
23:43 know the different systems and be
23:45
23:45 connected and as such or what we call
23:47
23:47 like essentially a knowledge graph.
23:50
23:50 And only then, when you have the uh
23:51
23:51 young people going in, they have the
23:53
23:53 right training, support, and tooling,
23:55
23:55 then they’re able to have
23:56
23:56 transformational change over many years.
23:58
23:58 >> Right. And and one of the things that
23:60
23:60 you’re able to do now with your
24:02
24:02 uh current company is help companies to
24:04
24:04 join the dots between all of those
24:05
24:05 different things, be it on their Slack
24:08
24:08 or within their intranet, and and help
24:10
24:10 them to navigate, I suppose.
24:12
24:12 Um which is powerful. I do want to come
24:14
24:14 back to saying to uh these qualities
24:16
24:16 that you’re looking for on [snorts]
24:18
24:18 um
24:19
24:19 being humble, uh being hungry.
24:22
24:22 Part of me is wondering if founders have
24:24
24:24 to be a bit disillusioned.
24:27
24:27 Cuz a lot of the qualities that you’re
24:28
24:28 talking about are about self-awareness
24:31
24:31 around um
24:32
24:32 being um
24:34
24:34 suitably um
24:36
24:36 sensitive and knowing what you’re going
24:39
24:39 into before taking a step forward. And I
24:41
24:41 can’t help but wonder if some of the
24:43
24:43 best founders are a little
24:44
24:44 disillusioned, and frankly, they use
24:46
24:46 ignorance, if I can use that word, as
24:48
24:48 their superpower. Yeah. So, disillusion
24:50
24:50 and ignorance is is is is
24:52
24:52 um
24:54
24:54 is I see it
24:56
24:56 I can see that in founders, certainly.
24:58
24:58 Disillusioned is something that I have
25:01
25:01 an issue with because
25:03
25:03 I’ll come to the answer in just a
25:04
25:04 second, but but there is a trend,
25:06
25:06 especially online, that talk about like
25:09
25:09 founders having always have had to gone
25:11
25:11 through trauma in their childhood.
25:13
25:13 And founders being like
25:15
25:15 just just
25:17
25:17 radically different to others. And
25:19
25:19 there’s a whole host of things which I I
25:21
25:21 I mean, I know many founders and this is
25:23
25:23 not not true. Most of my founder friends
25:25
25:25 have had incredibly privileged lives.
25:28
25:28 They frankly have had the best
25:29
25:29 education. And it’s it’s the the the the
25:31
25:31 level and quality I want for everyone in
25:33
25:33 in the in the country. And [snorts] so
25:36
25:36 they certainly you you can get
25:38
25:38 be well-rounded, you can be very
25:40
25:40 competitive, but you can get that in
25:41
25:41 sports on the weekends. You don’t have
25:42
25:42 to go through trauma. And it’s important
25:44
25:44 for people to know because I don’t want
25:45
25:45 parents who have ambitions for their
25:46
25:46 children to be entrepreneurs to to
25:48
25:48 somehow think that
25:49
25:49 >> [laughter]
25:49
25:49 >> that they’re disadvantaged because
25:51
25:51 they’ve they’ve because they’ve not gone
25:51
25:51 through any trauma.
25:53
25:53 Um and there is there’s uh founders
25:56
25:56 needing to justify, especially in the
25:58
25:58 world of AI, where you like the wealth
25:60
25:60 disparity is not just 10x 100x anymore,
26:02
26:02 it could be like a million x, right? Um
26:05
26:05 to somehow give this impression that
26:07
26:07 you’re Iron Man and like frankly like
26:10
26:10 very very different to others. You’re
26:11
26:11 you’re not. Most of these companies you
26:13
26:13 find a founder that has specific
26:14
26:14 strengths, certainly, often many
26:16
26:16 weaknesses. And they’re just have been
26:18
26:18 lucky enough to surround themselves with
26:19
26:19 just like incredible operators around
26:21
26:21 them. And together as a team, they fill
26:23
26:23 these gaps and they just go forward at a
26:25
26:25 time in a relative to to their
26:27
26:27 competitors doing a better job. And that
26:29
26:29 means they disproportionately return on
26:30
26:30 their time. They they sort of make it
26:32
26:32 make it happen.
26:33
26:33 Um
26:34
26:34 Now, there is a point around ignorance
26:38
26:38 and there is I feel, if you’re trying
26:41
26:41 something completely new,
26:43
26:43 you ought to it’s it’s some ignorance is
26:46
26:46 bliss, sure. And so
26:48
26:48 going in with a fresh pair of eyes,
26:50
26:50 that’s that’s why you kind of you’re not
26:52
26:52 often that’s because you’re not burdened
26:54
26:54 by biases from the past. Yes. And that’s
26:56
26:56 goes back to the Innovator’s Dilemma a
26:58
26:58 little bit in that in our identity
26:60
26:60 industry, the credit bureau world, um
27:02
27:02 they probably saw this is coming. I
27:04
27:04 mean, Kodak knew, there was evidence
27:07
27:07 that Kodak knew for well, digital
27:08
27:08 cameras are like the future. But you’re
27:10
27:10 making so much revenue, your customers
27:12
27:12 are happy, the whole organization is so
27:13
27:13 embedded to doing things today, it’s
27:16
27:16 extremely hard to then, frankly, pivot
27:18
27:18 or change or try something new. You’d
27:20
27:20 have to acquire a company, you’d have to
27:22
27:22 spin up a different team, you’d have to
27:24
27:24 maybe give them goals to develop an
27:26
27:26 innovative product that will probably
27:28
27:28 cannibalize your revenue. And you can do
27:30
27:30 that if, in my mind, if you it’s like
27:32
27:32 founder-led or a
27:34
27:34 fully empowered like CEO and not at the
27:37
27:37 behest of like shareholders or like
27:39
27:39 public markets.
27:40
27:40 >> Reporting quarter to quarter. Precisely.
27:42
27:42 Because this is uh
27:44
27:44 in those environments, it’s all about
27:46
27:46 what can you prove to me today and what
27:48
27:48 can I get a return on and evidence I
27:50
27:50 want certainty today, whereas these big
27:52
27:52 bets are like 10 years into the future.
27:54
27:54 >> Right. I’m going to ask you to pick a
27:55
27:55 number between 1 and 10. We’re going to
27:56
27:56 go a little more quick fire.
27:58
27:58 Pick a number between 1 and 10. I’ve got
27:60
27:60 some questions.
28:00
28:00 >> Six.
28:03
28:03 It’s a question about investors.
28:05
28:05 What do most investors misunderstand
28:08
28:08 most?
28:10
28:10 They
28:11
28:11 at an
28:13
28:13 early stage,
28:15
28:15 uh all they should really care about
28:18
28:18 is uh the team, which is certainly
28:20
28:20 founder, but those around the founder,
28:23
28:23 and the target, which is essentially the
28:24
28:24 the market opportunity. That is it. Do
28:26
28:26 not think of anything else. Uh
28:28
28:28 >> And what you’ve missed out there, or not
28:30
28:30 missed out, you’ve deliberately skipped
28:31
28:31 over what the business is actually
28:33
28:33 doing.
28:33
28:33 >> Yes, because in this world you kind of
28:35
28:35 define that broadly what your target is
28:37
28:37 very in very broad sense. That’s
28:39
28:39 pre-seed. And then at seed stage you
28:40
28:40 just add traction, the third T.
28:43
28:43 Um so, team, target, and then it’ll be
28:46
28:46 traction. And that in that order. No,
28:48
28:48 but that guy comes later. Right. Pick
28:49
28:49 another number. Four.
28:53
28:53 What’s a belief that you’ve changed your
28:55
28:55 mind on?
28:58
28:58 I have
29:00
29:00 uh I wrongly thought that if we hired
29:04
29:04 very experienced uh executives from
29:08
29:08 Meta, Google, Amazon, and others that we
29:11
29:11 ought to or at least certain I should I
29:13
29:13 outsource all the thinking and all the
29:15
29:15 decision like everything to to them. And
29:17
29:17 uh there are times where that like
29:18
29:18 really worked out and there are times
29:20
29:20 where it certainly did not. And now
29:22
29:22 because of AI, I recognize it’s because
29:24
29:24 they didn’t have the context or the
29:25
29:25 judgment or thing that just frankly
29:27
29:27 after 10 years would just be part of me.
29:29
29:29 And as such, I now certainly look for
29:33
29:33 experience and support and
29:34
29:34 brainstorming, but I don’t outsource uh
29:37
29:37 judgment-based or crucial strategic
29:39
29:39 decisions like I would have done sort of
29:41
29:41 7 or 8 years ago. Right. Really
29:43
29:43 interesting. Pick a final number. Two.
29:47
29:47 What’s a founder skill? We’ve talked
29:48
29:48 about this a bit. What’s a founder skill
29:50
29:50 that AI cannot replace? You have to I’m
29:54
29:54 all about companies even if they’re
29:56
29:56 called tech companies, they’re all about
29:57
29:57 people.
29:58
29:58 You have to go and sit across the table,
30:01
30:01 pay for the coffee, and try to persuade
30:03
30:03 an engineer or sales person whatever why
30:04
30:04 they should pick you.
30:06
30:06 And that is frankly charming them
30:08
30:08 because you you you want to them to
30:10
30:10 leave and buy into your vision. And
30:12
30:12 you’re pitching them an AI tool that
30:13
30:13 sounds the same as the half the other
30:15
30:15 people in the same coffee shop, right?
30:16
30:16 And maybe paying them less? And maybe
30:18
30:18 certainly will be. So, you’re asking
30:19
30:19 them to take a risk, which is like
30:20
30:20 equity. It’s like high risk, high
30:22
30:22 rewards. So, in 5 years if we end up
30:24
30:24 with nothing, you’ve you’ve just
30:25
30:25 sacrificed half your salary for those 10
30:27
30:27 years. But if it takes off, you’re 100
30:29
30:29 times, you know, in wealth terms more
30:31
30:31 than what you would have. So, it’s it’s
30:32
30:32 very much you you you’re buying into a
30:34
30:34 very high risk, high return. But you
30:36
30:36 can’t you can’t you can’t do that by our
30:37
30:37 emails. You can’t shouldn’t automate
30:38
30:38 that. You should sit with them human to
30:40
30:40 human
30:41
30:41 and uh
30:42
30:42 and essentially sell
30:45
30:45 division.
30:45
30:45 >> It’s fascinating. It’s what I’m taking
30:46
30:46 from this conversation is
30:49
30:49 some thinking about moderation. A little
30:51
30:51 bit like we humans would think about
30:53
30:53 exposure to the sun or to fire.
30:56
30:56 And it’s that moderation right through
30:57
30:57 from childhood. It’s also from a highly
30:60
30:60 technologically driven individual about
31:03
31:03 human human relationships.
31:05
31:05 Face-to-face time.
31:06
31:06 >> [snorts]
31:06
31:06 >> Um
31:08
31:08 let’s go much bigger picture.
31:09
31:09 When you think about the future,
31:12
31:12 what excites you? What keeps you up at
31:14
31:14 night?
31:15
31:15 Let’s go big picture as you like. I We
31:18
31:18 talked a little bit about the future
31:19
31:19 work. So, I study the future of work. I
31:21
31:21 healthcare, certainly. I I it’s
31:23
31:23 certainly thing it’s going to be the
31:24
31:24 breakthroughs that’s going to actually
31:25
31:25 touch people on and the mainstream and
31:27
31:27 they’re going to be like, “Okay, now I
31:28
31:28 understand. I’ve heard of AI. It’s
31:30
31:30 mostly been negative. I now, thank
31:32
31:32 goodness, I can get the scan and it’s
31:33
31:33 much quicker. I can do this and I can
31:35
31:35 get diagnosis on this.” So, that’s going
31:38
31:38 to touch people in a big way.
31:39
31:39 Education’s the same. Personalized
31:41
31:41 education, custom education specifically
31:43
31:43 to those who might have a specific
31:44
31:44 strength that they need to foster and
31:46
31:46 you can’t do that in a classroom of 30.
31:48
31:48 Or if you have a some sort of impairment
31:51
31:51 that could be a form of dyslexia,
31:53
31:53 autism, whatever. Like AI can absolutely
31:55
31:55 really change things for for for for for
31:57
31:57 those in particular. Um what worries me
31:60
31:60 is that back to the founder story, we
32:02
32:02 are AI by its nature works well at scale
32:06
32:06 in particular and will be so much
32:11
32:11 able to create a disproportion effect
32:14
32:14 around fewer startups becoming bigger.
32:17
32:17 And that’s
32:18
32:18 wealth concentration in a handful of
32:20
32:20 what otherwise [snorts] would be like
32:21
32:21 extremely well what end up being very
32:23
32:23 very powerful people
32:25
32:25 means this inequality is going to
32:27
32:27 continue. And unless there’s a
32:29
32:29 a thought-through redistributive layer
32:31
32:31 to that, you’re going to leave many
32:33
32:33 behind. Because I actually don’t think
32:35
32:35 there’s going to be unemployment. This
32:37
32:37 automation trend has been going on for
32:39
32:39 for a a long time, especially since the
32:41
32:41 like manufacturing automation that
32:43
32:43 started 40-50 years ago. What you’re
32:45
32:45 going to have is not an employment,
32:46
32:46 you’re going to have people pushed into
32:48
32:48 bad jobs. And there’s a book called BS
32:51
32:51 jobs. So you’re going to have people
32:53
32:53 moving into BS jobs and they’re not
32:56
32:56 going to be satisfied with that nor
32:57
32:57 should they be, especially when the cost
32:59
32:59 of living is like squeezing squeezing
33:01
33:01 every pleasure out of life. And that
33:02
33:02 that is certainly cause for concern. And
33:05
33:05 if this [clears throat] was a group
33:06
33:06 conversation with all of our previous
33:08
33:08 guests, one of them at least would say
33:10
33:10 that in previous revolutions, if you
33:12
33:12 like, you know, those car horses,
33:16
33:16 to use a well-known thought,
33:18
33:18 did lose their jobs.
33:20
33:20 You know, they didn’t go and retrain.
33:22
33:22 They didn’t go and do something else.
33:24
33:24 It was curtains for them.
33:28
33:28 If you have students coming off a school
33:30
33:30 conveyor belt which has led
33:33
33:33 meant they’re unequipped to take on any
33:35
33:35 of these jobs, that is a serious
33:37
33:37 unemployment issue looming.
33:40
33:40 Yes, and retraining makes sense.
33:44
33:44 Or we we didn’t succeed in that when
33:46
33:46 sort of the
33:48
33:48 UK industrial complex basically
33:50
33:50 collapsed in the ’80s. The American Rust
33:53
33:53 Belt, the auto manufacturing like that
33:54
33:54 collapsed. Again, retraining didn’t
33:56
33:56 happen. And this is called Middle
33:58
33:58 England and Middle America for a reason.
34:01
34:01 So
34:02
34:02 we’ve had recent examples of how like
34:03
34:03 the retraining didn’t happen.
34:05
34:05 And again, the challenge here for every
34:08
34:08 single role and every single function,
34:09
34:09 there’ll be a line drawn. That line will
34:11
34:11 mean if you’re proficient with AI,
34:13
34:13 you’re in all likelihood going to your
34:15
34:15 talent is going to grow significantly at
34:17
34:17 a greater pace than it has in the past.
34:19
34:19 If you’re below the line, you’re going
34:20
34:20 to fall behind. So this inequality per
34:22
34:22 role is going to increase. You already
34:23
34:23 have it in engineers, programmers,
34:25
34:25 computer engineers.
34:26
34:26 Junior roles, less less
34:28
34:28 you know, you can’t really find a job.
34:30
34:30 Mid-level and senior, especially senior,
34:32
34:32 you’re you’re earning more than you’re
34:33
34:33 more in demand than ever before.
34:35
34:35 Paralegals is the same.
34:37
34:37 Customer service is the same and beyond.
34:39
34:39 So, as such, I fundamentally believe
34:41
34:41 that we are going to have fewer people
34:44
34:44 in knowledge work doing high judgment
34:46
34:46 work being paid significantly more and
34:49
34:49 very many people, a greater proportion
34:51
34:51 and growing, will be
34:53
34:53 left to essentially do lower-paid work.
34:57
34:57 And actually, I’ve nothing wrong with
34:58
34:58 low-paid work. There’s a lot of
34:60
34:60 integrity and you know, I’m from a like
35:02
35:02 working-class background. When you talk
35:04
35:04 to someone working-class, like if
35:05
35:05 they’re on a low salary, they’re not by
35:07
35:07 default unhappy. The majority of the
35:09
35:09 world, by definition, is going to be on
35:10
35:10 a low wage or or like a below-average
35:12
35:12 wage, by definition. The problem they
35:14
35:14 have is the cost of living. Like you you
35:17
35:17 can be happily on £30,000 in Grimsby
35:21
35:21 in a house and like go on a holiday and
35:23
35:23 like have life and as long as you have a
35:25
35:25 dream of your kids doing better than
35:26
35:26 you. Most Most, I would say, are Well,
35:28
35:28 the problem now is they’re being
35:30
35:30 squeezed from all sides. They can’t
35:31
35:31 afford the health care. They can’t
35:32
35:32 afford the education. They can’t afford
35:33
35:33 the living. They can’t afford Their kids
35:35
35:35 are being
35:36
35:36 um abused and manipulated by social
35:39
35:39 media, big tech,
35:41
35:41 uh pharma. You know, like it just
35:43
35:43 everyone’s on the attack and they’re
35:45
35:45 they’re kind of eventually be like,
35:46
35:46 “Okay, this is not working. I want to
35:47
35:47 flip the table.”
35:48
35:48 >> So, on that, if you were
35:50
35:50 um
35:51
35:51 more involved in running the country,
35:54
35:54 what would you be doing differently?
35:56
35:56 There’s a So, I am a little bit of a I
35:60
35:60 think we’ve gone through like 20 years
36:02
36:02 of like win-win
36:03
36:03 uh
36:04
36:04 like let let Is there ways we can
36:06
36:06 structure the economy to maximize
36:08
36:08 win-win?
36:09
36:09 And I think we’ve we’ve kind of like
36:11
36:11 pushed that as far as it goes. I’m a lip
36:13
36:13 I’m I’m really into like a win-lose kind
36:15
36:15 of thing. I think concentrations of
36:17
36:17 power and wealth um won’t voluntarily
36:20
36:20 give that up. I think there have to be
36:22
36:22 like a loser scenario. And that means a
36:25
36:25 switched-on
36:27
36:27 uh ambitious
36:29
36:29 uh like
36:30
36:30 political organizers
36:31
36:31 that come in with the promise of taking
36:34
36:34 on big power and actually delivering
36:35
36:35 that. And what does that mean? Stronger
36:37
36:37 regulation.
36:39
36:39 You know, all utilities, I mean most of
36:41
36:41 the things that you look at in the UK,
36:42
36:42 they’re not trending in the right way.
36:43
36:43 >> [clears throat]
36:44
36:44 >> All though there’s course correction to
36:45
36:45 be done there. And secondly, the safest
36:47
36:47 thing you can do on an education,
36:49
36:49 education, education is investing in the
36:51
36:51 under 18s. And
36:54
36:54 course correcting or frankly reversing
36:56
36:56 many of the things that have happened.
36:58
36:58 We’re closing social
36:59
36:59 clubs which after school libraries,
37:02
37:02 classroom size are growing. Outside of
37:04
37:04 London, the country trajectory is like
37:06
37:06 really awful and many parts of London is
37:08
37:08 pretty awful, too. As such, let’s invest
37:10
37:10 in under 18s properly for for like the
37:12
37:12 future and let’s rebalance a cost of
37:15
37:15 living crisis and everything else that
37:17
37:17 goes with it.
37:17
37:17 >> So, a listener I have to say
37:20
37:20 AI uniquely for the, you know, is a
37:23
37:23 technology that can actually really
37:24
37:24 deliver that in many in many areas as
37:26
37:26 long as it’s managed properly and
37:27
37:27 regulated properly. Okay, the
37:29
37:29 I think the ambiguity is still you’re
37:31
37:31 now schools minister.
37:33
37:33 How techy or how how tech
37:36
37:36 enabled do you want our schools to be?
37:39
37:39 So, the first thing is you got starting
37:41
37:41 with basics. So, why don’t I start by
37:43
37:43 saying like what what the what the win
37:45
37:45 could look like. If I if we say founders
37:48
37:48 and operators show high capable, we want
37:51
37:51 more of them.
37:53
37:53 Then if I take as a proxy probably
37:55
37:55 school educated um
37:58
37:58 sort of founders take up 75% of seed
38:01
38:01 stage funding.
38:03
38:03 And they only represent 7% of the whole
38:04
38:04 population. So, I would want that level
38:07
38:07 of education and standard for for
38:08
38:08 everyone in the nation. So, by
38:10
38:10 definition, if you give the same level
38:12
38:12 of education as a private school in the
38:13
38:13 UK, you’re going to take
38:16
38:16 10x increase in the number of founders
38:17
38:17 and operators around them.
38:19
38:19 So, that to me is the number one thing
38:22
38:22 to increase the tech stock of the
38:24
38:24 country is to give everyone like a gold
38:26
38:26 standard education. So, back to
38:30
38:30 education minister, the first is like we
38:32
38:32 have a child hunger poverty problem,
38:34
38:34 right? I I you know, the two child
38:36
38:36 benefits cap being lifted is is
38:37
38:37 wonderful, but
38:39
38:39 essentially nutrition goes goes without
38:41
38:41 saying. Secondly, classrooms with
38:44
38:44 teachers that are well paid and that
38:45
38:45 they have the tooling and resources they
38:47
38:47 need to actually move the needle. The
38:49
38:49 third thing is
38:50
38:50 after school clubs and everything around
38:52
38:52 that. I mean, me growing up at the age
38:53
38:53 of 10 when we moved to the UK, I got
38:55
38:55 like milk. I got I like Mrs. Salvaterra,
38:57
38:57 I remember her name. She spent an hour a
38:59
38:59 day helping me with like improving my
39:01
39:01 English.
39:02
39:02 After school, I played table tennis one
39:05
39:05 one one after school. And and all So, we
39:07
39:07 were in a poor neighborhood at a poor
39:09
39:09 time, but I actually felt
39:11
39:11 that I wasn’t missing much. Later, I got
39:13
39:13 a grant and a bursary to go to Oxford. I
39:15
39:15 never really struggled with funding at
39:17
39:17 all.
39:18
39:18 And I was just helped throughout the
39:19
39:19 way. I I I I’d look at life as like
39:21
39:21 snakes and ladders, and all I could
39:23
39:23 experience was ladders. Probably very
39:25
39:25 honest So, we can get there and I had
39:27
39:27 that not long ago. It wasn’t It wasn’t
39:28
39:28 like 50 years ago. I just had it 15
39:29
39:29 years ago.
39:30
39:30 I will say though, my nephews, cousins,
39:32
39:32 or those who I visit now, they have the
39:34
39:34 same board of snakes and ladders, but
39:35
39:35 it’s just filled with snakes and very
39:38
39:38 few ladders.
39:39
39:39 So, let’s
39:41
39:41 finish the conversation where some
39:43
39:43 conversations about your life would
39:45
39:45 start, which is in those early years.
39:48
39:48 You weren’t born in the UK. How did
39:51
39:51 those early years affect the person you
39:53
39:53 are today? Yeah, I actually was born in
39:56
39:56 Manchester.
39:57
39:57 >> me. Now, now there’s there’s a twist
39:58
39:58 here. Yes. By the age of one, we moved
40:00
40:00 to Iran. So, my mother’s English and my
40:02
40:02 father’s Iranian.
40:03
40:03 Age of one, moved to Iran until I was
40:05
40:05 age of 10, and then we moved back to to
40:07
40:07 the UK. When when we moved back in part
40:10
40:10 because
40:11
40:11 my mother was accused of being a spy.
40:13
40:13 But in Iran, it’s like very common to be
40:14
40:14 accused of So, it’s It’s not not a big
40:16
40:16 deal. So, meaning I got to move around
40:18
40:18 quite a lot, went to different schools,
40:20
40:20 different like places, and
40:22
40:22 um I think you at a young age
40:25
40:25 when you have these different
40:26
40:26 experiences, you do develop certain
40:28
40:28 things uh including being comfortable
40:30
40:30 with change. I think in a startup
40:32
40:32 environment, most founders that I’ve met
40:34
40:34 and seen, they are somewhat more
40:36
40:36 comfortable with I don’t want to call it
40:37
40:37 chaos, but in some ways change,
40:39
40:39 uncertainty, and so on. And I certainly
40:42
40:42 have had colleagues who really we need
40:44
40:44 to create an environment for them which
40:45
40:45 is not a lot of variance. Like they need
40:48
40:48 to be solid, you know, they can come at
40:50
40:50 9:00, ship code, do whatever, leave at
40:53
40:53 5:30, whatever it is. And yeah,
40:55
40:55 different people need different types of
40:57
40:57 environments where they where they do
40:58
40:58 well. Yeah, and that being comfortable
41:00
41:00 with change and being uncomfortable
41:02
41:02 >> If you’re not like that, you should not
41:04
41:04 do a startup.
41:05
41:05 You really should not.
41:07
41:07 Do you think your current startup would
41:08
41:08 hire you?
41:10
41:10 Historically, one is told not to hire
41:13
41:13 founders.
41:14
41:14 And I think uh truth be told, if I had a
41:17
41:17 clone of mine to hire, I would say do
41:19
41:19 not hire him. Why?
41:20
41:20 >> Um because I feel like early-stage
41:23
41:23 startups, it should be crystal clear on
41:25
41:25 messaging and like one vision and one
41:28
41:28 captain of the ship.
41:30
41:30 And then everything else falls into
41:31
41:31 place. You have um if there’s no
41:34
41:34 alignment, if it was if it was a literal
41:37
41:37 clone of mine, then that’d be heavenly.
41:39
41:39 But [snorts] if it was like a twin that
41:40
41:40 had a different upbringing and different
41:42
41:42 assumptions like and grew up in a
41:43
41:43 different part of the world, uh you you
41:45
41:45 create kind of like two opinions in a
41:47
41:47 company that are two strong voices. And
41:50
41:50 that I don’t think that’s healthy. You
41:51
41:51 should have complementary skill sets,
41:52
41:52 not identical ones.
41:53
41:53 >> Right. Understood. Final
41:55
41:55 uh quick fire. We talk a lot of Future
41:57
41:57 Forum about leaving your lane,
41:59
41:59 uh about learning from other industries.
42:01
42:01 Any industries you have your eye on that
42:03
42:03 you’re conscious that you’re pollinating
42:05
42:05 ideas in from?
42:07
42:07 I can say I mean the education and the
42:10
42:10 training industry in particular cuz I’m
42:13
42:13 all about how does a human skill improve
42:15
42:15 to be able to properly unlock AI? I look
42:17
42:17 at the late ’90s, early 2000s. Those
42:20
42:20 were in finance and able to master
42:22
42:22 Excel, you know, Microsoft Excel, got
42:24
42:24 the best finance jobs. AI is going to be
42:26
42:26 the same in every single role,
42:28
42:28 knowledge worker and high judgment in
42:29
42:29 particular. Those who are now investing
42:31
42:31 in learning AI are going to become
42:34
42:34 essentially the future of what finance
42:35
42:35 was in the 2000s. And I we are very much
42:39
42:39 like a tech and software company, but we
42:40
42:40 partner with training providers and
42:42
42:42 those who are specialized in improving
42:43
42:43 human skills. So that by nature of my
42:46
42:46 work, I’m actually very active with with
42:47
42:47 that industry, too. I actually came from
42:49
42:49 the learning conferences technology just
42:50
42:50 last week. Yeah, and I’ve seen you as a
42:53
42:53 company bring people together to think
42:55
42:55 about this as well.
42:57
42:57 Final quick fire question. You have been
42:60
42:60 closely involved with the launch, in
43:02
43:02 fact, I think of you as the founder of
43:04
43:04 the London AI hub.
43:05
43:05 So in a world of technology, why do
43:07
43:07 physical spaces matter in that sense?
43:09
43:09 >> I’d say more than ever because of
43:11
43:11 essentially learning from each other in
43:13
43:13 fast-paced change space. So I
43:16
43:16 in the early sort of 2014,
43:19
43:19 2015, spent so much time at Level 39 in
43:21
43:21 Canary Wharf. That was where the
43:22
43:22 FinTechs were. You were there all the
43:23
43:23 time. And
43:25
43:25 Revolut were tenants there and the
43:27
43:27 Financial Conduct Authority would do
43:28
43:28 events there. So I just saw that us
43:30
43:30 being together, including the regulator,
43:32
43:32 policy, and everyone, it led to London
43:34
43:34 becoming the FinTech capital. And so
43:36
43:36 when I wanted to move to the central
43:38
43:38 London
43:39
43:39 18 months ago, I wanted to go to the AI
43:41
43:41 hub, essentially where all AI companies
43:42
43:42 would be building and learning from each
43:44
43:44 other. It didn’t exist. And as such, I’m
43:47
43:47 very lucky that both Merantix, who are
43:48
43:48 based in Berlin, and Founders Forum,
43:50
43:50 they both had similar ideas. So we kind
43:52
43:52 of grouped together and we’ve been very
43:54
43:54 lucky ever since.
43:56
43:56 Why is the UK,
43:58
43:58 and I’ve heard you say this on multiple
43:59
43:59 occasions, so it is an assumption in the
44:01
44:01 question. Why do you think the UK is a
44:03
44:03 great place to build something?
44:05
44:05 Ultimately, it’s all about where the
44:07
44:07 talent is. Now, historically, human
44:09
44:09 capital mattered, but financial capital
44:12
44:12 mattered a little bit more. To build a
44:14
44:14 billion-dollar company, you needed,
44:15
44:15 historically, $300 million.
44:18
44:18 And the US has four times the amount of
44:19
44:19 financial capital than all of Europe.
44:21
44:21 So, raise the most money, hire the best,
44:24
44:24 uh and then you you sort of end up more
44:26
44:26 more often than not building the best
44:28
44:28 technologies. I would say AI, uh
44:30
44:30 especially in the application layer, um
44:33
44:33 we are now seeing, you know, 2 3 years
44:35
44:35 in, is costing a fraction of what it
44:37
44:37 used to. Probably a quarter, if not
44:38
44:38 less. Which means the financial capital
44:40
44:40 that the US has is like loses its edge.
44:42
44:42 And then you default to human capital.
44:45
44:45 And as such, Europe has just such
44:47
44:47 amazing talent, especially like London,
44:49
44:49 we’re surrounded by amazing universities
44:51
44:51 and all this amazing talent. So, now
44:52
44:52 that financial capital doesn’t matter as
44:54
44:54 it once did, there’s ample funding here.
44:57
44:57 I’d argue there’s a far too much
44:58
44:58 financial capital. But anyway, what the
44:60
44:60 the bottleneck is human capital. And you
45:02
45:02 basically are able to build amazing
45:04
45:04 talent here, and they stay, and they
45:06
45:06 they essentially uh with AI, you can
45:09
45:09 build a company here and take it global.
45:11
45:11 So, if we think about the future, I can
45:14
45:14 imagine you doing so many different
45:15
45:15 things. What do you think you’ll end up
45:17
45:17 doing? I would really want to build Auto
45:20
45:20 for the next 20 years. And I tell all my
45:22
45:22 investors, like, you know, if you’re
45:24
45:24 investing, just bear in mind this is not
45:26
45:26 like a quick thing. This is like a
45:27
45:27 20-year This is a slow path option. Um I
45:31
45:31 have also It would be my last startup.
45:33
45:33 So, this is my second and last one. Um I
45:37
45:37 don’t even know or have thought about
45:39
45:39 what I might do in 20 years from now.
45:41
45:41 >> All right. Well, now’s the chance. I’m
45:43
45:43 going to clone you on the spot. Hussain
45:45
45:45 Kasai one is going to build Auto for the
45:48
45:48 next 20 years. What’s the second one
45:50
45:50 going to do? I’d probably send him off
45:53
45:53 far away.
45:54
45:54 >> [laughter]
45:54
45:54 >> But I’d take him to like developing
45:56
45:56 countries, whether it’s India or
45:57
45:57 Nigeria, to see how AI could be used to
46:00
46:00 close the gap to your typical 18, 20,
46:03
46:03 22-year-olds
46:05
46:05 uh to be able to essentially develop and
46:07
46:07 and deliver services that his startup
46:10
46:10 had not been able to.
46:11
46:11 What a fascinating answer. I want
46:12
46:12 another hour to unpick that. Um
46:16
46:16 final question. And thank you so much
46:18
46:18 for being our guest. It’s been
46:19
46:19 absolutely fascinating. I would have
46:20
46:20 talked for another hour at least or
46:22
46:22 >> [snorts]
46:22
46:22 >> listen for another hour.
46:23
46:23 Um we ask all our guests to send a memo
46:26
46:26 to the future. Could be a version of
46:28
46:28 your future self. It could be to a
46:29
46:29 future founder, maybe 10 years time. Uh
46:32
46:32 it could be 20 years time. You need to
46:34
46:34 look at them uh look into the future and
46:36
46:36 tell us what you’d say.
46:40
46:40 In 10 years time
46:42
46:42 I hope that you are
46:44
46:44 bright enough to know
46:46
46:46 >> [laughter]
46:46
46:46 >> that when all is said and done, all that
46:48
46:48 really matters is family and health. Uh
46:51
46:51 well above everything else by far.
46:54
46:54 Uh second to that, if you have been uh
46:58
46:58 able and lucky enough to build a special
46:59
46:59 company you still recognize that it’s
47:02
47:02 built as a result of the team and that’s
47:04
47:04 the number one marker beyond customers
47:07
47:07 liking the product is whether team
47:08
47:08 members leave having developed far more
47:11
47:11 than they would have had they done a PhD
47:13
47:13 or been at another company.
47:15
47:15 And that they are inspired and uh to
47:17
47:17 build their own companies. But crucially
47:19
47:19 to build companies that are known for
47:21
47:21 their amazing culture uh beyond the
47:24
47:24 products and and and so forth cuz
47:26
47:26 um as you know, given that you’re me uh
47:29
47:29 you just build a special culture and the
47:31
47:31 rest kind of follows. So, yeah. In 10
47:33
47:33 years when you when you listen to this,
47:34
47:34 I hope this will resonate.
47:37
47:37 That is a phenomenal memo. Uh it’s been
47:40
47:40 absolute pleasure uh talking to you,
47:41
47:41 Hussein Kassai. Thank you so much for
47:43
47:43 joining us. Real pleasure.
47:46
47:46 So, that’s our guest today, Hussein
47:48
47:48 Kassai, a founder and someone I thought
47:50
47:50 we were going to have a conversation as
47:52
47:52 in many ways we did about building
47:53
47:53 organizations and uh enterprises and
47:56
47:56 instead we’ve ended up talking, haven’t
47:58
47:58 we, about relationships, about
47:60
47:60 human-to-human connection. Also right
48:02
48:02 through from education
48:04
48:04 through to family, but also that cloning
48:07
48:07 question, that idea that there are so
48:09
48:09 many challenges out there which could be
48:10
48:10 solved and for many founders could
48:12
48:12 distract them. And that brings us back
48:15
48:15 to that power of a singular focus on a
48:18
48:18 particular mission in hand. So I hope
48:20
48:20 you’ve enjoyed today’s conversation as
48:21
48:21 much as I have. Please do share it with
48:24
48:24 others if you haven’t. Ask who else
48:25
48:25 you’d like to hear from. It’s tomorrow.
48:28
48:28 It’s the Future Forum podcast and I’m
48:30
48:30 Ollie Barrett. Thank you for tuning in
48:32
48:32 and until next time, goodbye.
All episodes
