May 1, 2024 • 47 mins
Have you ever wondered how a self-described academic underachiever can morph into a powerhouse entrepreneur? Dr. Alex Allen, co-founder of Kortical joins us to share his transformation from a laid-back student to a PhD holder and AI innovator.
His insights are a treasure trove for any budding founder, emphasising the significance of knowing your 'why' and the symbiotic relationship between consistent progress and mental well-being.
What does the future hold for AI? We uncover the prospective role of AI agents across a variety of industries and wide - reshaping how we handle tasks from travel planning to legal research.
We don't just talk shop; our chat takes a turn towards the personal challenges and growth experienced along the entrepreneurial journey. From recognising the importance of goal setting to engaging in disciplines like boxing, Alex shares how confronting fears and stepping out of one's comfort zone are critical for self-improvement. He recounts how these experiences have been instrumental in him becoming a more resilient person, both in and out of the ring.
You can connect with Dr Alex via LinkedIn or check out Kortical.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
What piece of advice have you been imparting to other
younger founders or peoplestarting out over the last few
years where you've learnedsomething and you think you know
?
I wish I'd known this.
Speaker 2 (00:12):
One would be trying to ask yourself like why am I
starting this business?
What am I actually trying toget out of it?
Like, are you trying to changethe world or do you want to make
money?
Do you want to provide for yourfamily and have a comfortable
life?
Because those take radicallydifferent paths.
Speaker 1 (00:27):
Alex is a chief data scientist and the co-founder of
Cortical.
Cortical is the automatedmachine learning platform for
professional data scientists.
Speaker 2 (00:38):
I think, anything where you can see improvements
in yourself or some other aspectof your life that can give you
a, I think, fundamentally humanpsychology.
If you see growth somewhere,that's an incredibly nourishing
thing, a sense of being on planand going towards a purpose, and
even if you're making smallprogress, that's quite a
powerful thing for mental healthand sense of well-being.
(00:59):
So if you're, you know, at theend of each week you're like,
yes, I've got a bit closer to mygoals.
Speaker 1 (01:05):
That's superb advice for people listening.
Hey everybody, it's GregSheehan.
Welcome to my podcast, whereyou will hear from a range of
guests, including those from thestartup world and those that
have had incredibly interestinglives and some stories to tell.
I would really appreciate it ifyou could hit the follow button
and share this amongst yourfriends, but, as you know, time
(01:27):
is limited, so let's get on withit and hear from our next guest
.
My guest today is Dr Alex Allen.
Alex is a chief data scientistand the co-founder of Cortical,
and we will get into that as theshow goes on.
Cortical is the automatedmachine learning platform for
professional data scientists.
Welcome to the show, alex.
Great to be here.
(01:47):
Thanks for having me.
It's a real pleasure to haveyou here.
We were introduced by somebodythat believes very strongly in
you and said look, I needed tointerview you.
In fact, she was quite adamantthat I needed to interview you.
She just absolutely loves whatyou do and very impressed by you
.
I generally start these bytalking a little bit about your
origin before we start to getinto Cortical and the story
(02:08):
around that, but I'd love toknow what sort of early
upbringing you had.
Were you always that kind ofentrepreneurial type?
Were you always a math-y kid?
What's the origin story?
Speaker 2 (02:17):
for you.
Yeah, that's a good one.
I mean I think I went to.
I grew up in Essex, which isjust sort of east of London, so
it's about a 50-minute drivefrom London, but it was
relatively rural.
I went to a comprehensive schoolwhich is like a state school,
and I was like it's kind of likefairly academic, but one of
those people that never reallyworked hard enough, if you know
what I mean.
I do, and I did things likecomputer science and business at
(02:40):
college.
And then I was looking fordegrees you could do and it was
like one degree in the country.
It was called cybernetics andartificial intelligence.
I just thought that soundsincredibly badass.
I'm just gonna have to do that,as I think that was.
This is in 2006, so you knowthis is way before AI was really
kind of a big deal.
But I was always a big fan ofscience fiction and I've always
(03:02):
thought that you know AI isgoing to be like it is in the,
in the movies.
You know when you're talking toa robot or something like from
Star Trek.
So I was always keen to kind ofget in on it and at first at
uni I was.
I struggled a little bitbecause I think there's a level
of oversight you get at collegeand things which you just at uni
you can just like you don'thave to go to lectures.
So I ended up kind of goingthrough some personal events
(03:24):
like my father passed away andthings like that, which kind of
led me to take a few monthsbreak.
And then I got into kind ofreading like self-help books and
they range from incrediblytrite to sometimes quite
profound, and through all ofthat I kind of came up with a
kind of personal philosophyaround setting myself goals and
targets and holding myselfaccountable.
(03:46):
And then I ended up just beingsomeone that worked really hard,
which is kind of I don't knowif it sounds realistic, but it
turns out you can changeyourself from someone that's you
know, never that I could, youknow self-control or
self-discipline into someonethat does have it.
And I found, when I was able tokind of consistently apply
myself and I think consistencyis like the really key thing
that I was doing a lot better.
(04:06):
And my third year at uni that Iwon a scholarship to do a PhD in
what was called data miningback then, but you'd probably
call it data science or AI now.
It's probably the same thingand that went well.
And, yeah, I was kind of duringduring the course of my life
I've always been obsessed withtrying to start a business and
did various things like fromselling memory phone mattresses
on e-commerce solutions to, youknow, building websites for
(04:28):
people and things like that.
So, kind of towards the end ofmy PhD, I just started doing
consulting in London.
I just said, yeah, I'm an AIconsultant.
And then, you know, because noone was doing it and and
actually I did legitimately haveexperience in a field that you
know this is this is back in2014, 15, there was really no
one else doing it and I was ableto do some cool stuff with some
(04:49):
interesting companies, whichgave me like a reasonably
interesting portfolio.
By the time I'd sort ofgraduated uni sorry, my PhD,
really and then I was obsessedwith trying to start a tech
business and I ended up bumpinginto a chap was giving a giving
a talk at a conference and therewas this fellow at the back
drinking, drinking the free wine, and I was like he's probably
(05:10):
my guy.
So me and him got chattingduring the rest of the talks and
it turns out he was like anex-McKinsey.
He did sort of systemsoptimization, but he did it just
for himself and we got onreally well and he was a lot
older than me, like 25 yearsolder, but we we were like,
right, cool, let's start abusiness together.
And then we kind of there was afew other people in my circle
and then we, we all got togetherand started this, this business
(05:30):
which was around.
It sounds it was kind of like aboring application domain, but
the technology was quite cool.
We sort of took two days out andsort of brainstormed what we
could do and we came up with theidea of but, like call centers,
right, you have to go through alist of leads and and call them
and they just randomly assignthem to agents.
Um as well, what if we couldlearn about what an agent might
(05:53):
be good at in terms ofdemographics?
So let's say you're you know,I've got a fixed scottish accent
you might actually resonatebetter with people at the north
of the country, for example, andthings like time of day, like
if you're calling someone whoworks in the city at like three
in the afternoon they'reprobably not going to be in, but
if you're calling someone thatlives in the suburbs, there's a
chance, there's a you know, anew parent or someone that's
(06:13):
going to actually answer thephone.
So we were like, okay, we couldoptimize the time of day and
the right agent, then we cancreate something that could
basically, just throughallocating these leads
differently, create an an upliftin sales, and we started
working on the technology.
Really.
It was at that point I kind ofbumped into really my current
co-founder and that wascompletely random.
(06:35):
It was like our mutualex-partners invited us both to
this bonfire night.
They weren't ex-partners at thetime, so it wasn't something
weird like that and we juststarted talking about neural
networks and ai stuff.
And he was a.
He was working at barclaysdoing big distributed kind of
machine learning things, youknow big investment banks.
So, okay, cool, this guy'slegit um, and I've eventually
(06:56):
invited him to join this, thisnow quite large group of
founders there's like five of usand then we kind of first we as
we were building the technologyand we were going to try.
You know we got a few trialswith a few call centers and it
went really well.
But we kind of found that themarket was like a terrible one
because these, these operationsoften run on the knife edge of
profitability, so they're notreally keen on dropping 50 grand
(07:19):
a year on some untested machinelearning solution which no one
believed in.
No one knew what ai was, no oneknows what machine learning was
.
You'd have to start every pitchwith like a 20 minute sort of
educational session on.
I know machine learning isgoing to be really big, it will
really help you.
And these are kind of like hardbits and sales people.
They're like I don't like this,I don't know what this is, what
are you even trying to sell me?
(07:39):
We're like we'll just do it forfree then.
Yeah, but yeah, but you know,to do it for free we'll need to
integrate into your coolingsystem.
And it wasn't really like a niceeasy path to demonstrating the
technology.
And so we were trying to raisesome money to looking to VCs and
angels and a bunch of them weresaying like well, the kind of
technology we built to build themachine learning itself, we'd
(08:00):
kind of ended up.
We ended up realizing thatevery single time we had to do
one of these POCs, we had tokind of build a totally new
machine learning model fromscratch, because the product
will be different, thedemographics will be different,
might be similar sort of datainputs, but realistically you're
going to need maybe a totallydifferent type of model and
certainly a different set ofmodel parameters.
So we started building a kindof automated machine learning
(08:22):
framework to kind of speed thesethings up.
And that got to a point whereit was a kind of automated
machine learning framework tokind of speed these things up.
And that got to a point whereit was actually kind of fairly
advanced and these vcs were likewhy are you not just selling
that?
Why are you trying to sell thecall centers?
Um, this was before somethingcame into force called gdpr,
which is the uk's kind of datalegislation.
I think it was originally aeuropean thing and now it's just
(08:43):
the uk's, but I think it's asimilar thing to happen in
Europe, and that basically meantyou couldn't cold call people.
I mean, you still get loads ofcold calls now, but it's
technically illegal.
So isn't what you'd call anideal market, one that's
effectively being legislated outof existence?
And so me and my currentco-founder decided to kind of
leave this.
We kind of brought everyone out.
(09:04):
It was kind of pre-revenue, soit wasn't a huge the company, it
wasn't like we needed to investmillions in that and we started
doing basically taking intechnology, because we brought
the other guys out, we had thetechnology.
I mean me and him had built thewhole thing more or less anyway
, and this was 2016.
And then I mean we had a fewcontacts and a few people we'd
(09:25):
met.
And again, this was back when,if you Googled AI consultancy
London, we were the top resultwith basically no marketing
spend, just because we were theonly people saying we did it.
And we ended up landing ourfirst job, which was working
with another consultancy akin tosomeone like Accenture it
wasn't them um, to basicallybuild the AI roadmap for one of
(09:50):
the UK's largest banks.
So it was like a huge projectto build this whole massive AI
roadmap.
It was basically our first realjob, which is, you know to, by
today's standards, are likeutterly insane.
Um, you know really hard andyou know put a lot of hours in
and ultimately it was consideredreasonably successful and then,
you know, everything kind ofsprang off the back of that.
(10:11):
So, yeah, that was kind of theorigin story.
Really, I kind of veered moreinto the company there, but it's
the two, that kind of I guess.
Speaker 1 (10:18):
I'm so used to giving this talk.
Yeah, no, it's perfect.
And so you get started and youwin a bank as a customer.
I mean, how would you describewhat Cortical is now Like if you
were standing there and you had30 seconds to pitch what
Cortical does?
What does it do?
Speaker 2 (10:35):
We reduce the cost, risk and time of going from a
data and an idea into somethingthat lives in production in the
real world.
That's making you money usingmachine learning.
That would be the very, veryelevated pitch.
I mean, what do we actually do?
Well, we've got a platform thatwe've built over.
The same kind of engine that Iwas talking about from the
previous company is still thecore of our product and we have
(10:58):
a kind of mix of clients likecharlotte tilbury fashion brand.
They use us as really aplatform that they build things
on themselves.
Other companies such asDeloitte.
We've been heavily invested inhelping them, working with them
to kind of build the product aswell.
It's still underlined by ourtechnology.
So we kind of got a bit of amix of self-serve and more
(11:19):
consultancy slash kind of pointsolution led customers.
But ultimately, yeah, we buildeverything on our own technology
and as a result, we do things alot faster and we can do POCs
for a lot quicker.
And you know we can often dosome of the initial work
basically for free because it'sso quick for us to do.
So you know, a normal datascience project, like if you
(11:41):
were to come to me and say, ohokay.
So I want you to try andpredict the views for my podcast
.
You know, going forward,depending on who you go to, that
could be like a three-month bitof work to kind of come up, you
know, with an exploration phaseof data, a kind of data
analysis phase, initial modelbuilding, and these things are
all done manually and they allrequire kind of iteration and
(12:03):
like experimental feedback loops, and we can because our
platform kind of closes a loopon a lot of these things we can
kind of do that very, veryquickly.
So we could probably, in like aday or two after getting the
data, have some idea of howpossible that is and what kind
of accuracy we might get.
So we've been, as a result, beenable to kind of take projects
where I mean the normal successrate for ai projects, isn't it,
(12:24):
and by the success I mean takingsomething from someone's idea
and actually turning that intoproduction.
It's like less than 10according to gartner.
We're in like the 90 region ofsuccess, and that's really
because we take only take thingson.
We think we're going to be ableto do.
We are able to, because of ourplatform, do a bunch of the
initial kind of sniffing out of.
Is this viable for as part ofthe pre-sales process?
(12:47):
So when we get to that kind ofyou know actual project, we've
got confidence that we're ableto deliver it.
Speaker 1 (12:53):
And is it a mix of a services business and a software
business?
Do you do consulting around theoutside of this as well?
Speaker 2 (13:01):
Yeah, the mix of consulting to just SaaS has been
veering in the direction ofSaaS, which is good because
that's more scalable, but Ithink ultimately we're still in
a market where most companiesaren't doing really anything in
production with AI.
Some companies are A lot havestarted to think about it, but a
lot are still unsure as towhere to get started.
(13:23):
So consultancy is a part ofhelping companies through that
process and we've got eightyears of experience of doing
that now, which is quite a lotin such a new field.
So we do have a lot of value tooffer there.
And when we started thisbusiness, when we first came up
with this idea for the platform,there was annoyingly although
we didn't know about it at thetime one American company called
(13:44):
Data Robot that was alreadydoing it.
This is always one that getsthere first and you know, we
completely independently came upwith the same idea and over the
course of lifespan, google,amazon and Microsoft have all
released equivalents to whatwe've got, which makes the sales
process more challenging.
Right, because these companieshave already got big,
longstanding relationships withthese large, trusted providers
(14:08):
and they've got all through thedata security and it's like oh,
you want to enable this elementof Google Vertex, for example.
You want to enable that.
That's just.
You've already got that throughyour data security, you've
already got it through your ITops and whatever.
You just tick a few boxes andit's there, whereas we've got to
say, look, no, we'll go throughall that process.
So us bringing that personalelement.
(14:29):
So if you go within those bigplayers bringing that personal
element, so you know, if you gowith any of those big players,
they're not going to give youanywhere near the level of
handholding and help that you'llget.
And if they are, if they, ifyou do pay for the consulting
package, it's not going to bethe founder or anyone close to
that or be, you know, relativelyjunior person who's going to
have a lot less experience thanus.
So that's, it's become a pathfor us to differentiate
(14:50):
ourselves and to provide a kindof better quality of service.
But we are seeing as morecustomers just realize they need
things like what we're doing,that we have to do less of that,
which is good, because it'sharder to scale a consultancy,
obviously, and keep the qualityhigh.
Speaker 1 (15:06):
It's interesting because competing with the big
players now in AI must be justalmost impossible to take them
on at their own game.
So to personalise it and, asyou say, be the founder who's
potentially leading some of thestrategic conversation, is the
point of difference for you.
So just give us a bit of asense of what does the company
(15:27):
look like now, to the extentthat you're willing to share.
So you started with the two ofyou, and then what are we?
Eight years on?
What is the company?
Sort of shape does it have now?
Speaker 2 (15:35):
Yeah, so after two years of starting the business
we raised $1.2 million.
This was from angels and thenwe scaled the team to like 15,
16 people and I think me and myco-founder are both really good
at building their technology andbuilding DOCs and doing their
sales, which is kind of a nicemix as a starting team.
(15:57):
But one of the things, whilewe're good in front of people
and we're good at selling theproduct, that doesn't
necessarily translate to beingable to scale a sales team and
find a product, understandproduct market fit and there's a
whole bunch of stuff that, as afounder yourself, you probably
know about the journey there.
Right, it's not easy and whenyou get a huge wad of investment
(16:19):
you've got some chance to testand learn and make some mistakes
.
But you know, ultimately thereason most businesses fail is
because there's a learning curveand a runway and you've got to
sort of get to that point in thelearning curve while you still
have some money left and wemanaged to do that.
But we also had a huge, Iwouldn't say catastrophe.
(16:40):
But you know, when COVID hit wehad about 50% of our revenue.
That was, you know, we prettymuch closed the deal but not
signed a contract for a largeset of really big projects and
then, as soon as covid hit,every single business was like,
yeah, no, innovation budget,we're gonna hunker down.
So we ended up with a big holein our cash flow so we had to
(17:03):
scale the team back a little bit.
Um, so we're now about 10people, but we've kind of got to
the point where we don't have arunway, we've self-sustaining
on our own revenue, which is amuch, much less stressful place
to be, and I'd probablyrecommend getting to that place
as soon as possible as abusiness owner, because you're
just then able to do moreexperimentation.
And we've now got to a placewhere we're much closer, closer
(17:26):
and or have got some realevidence, I think, of product
market fit, which is somethingwe, you know, we still, we still
closing deals, but it was kindof all quite eclectic and they
weren't, you know, it was more,maybe, a product of the nascence
of the market, rather than wefound something solidly we could
repeatedly win at.
And I think, you know, thefrustrating thing for us was we
(17:47):
were always like three stepsahead in terms of what we wanted
to build, versus someone likeMicrosoft, and we knew exactly
what they were going to buildbecause we were thinking well,
this is obviously the next stepand they would have a team of
like 50 or 100 people buildingtowards it and we could spare
like one developer.
So well, we had the ideas first, the execution often we can get
a basic version out, but theneverything they release would be
(18:10):
already integrated into a wholesuite of other products.
So it just became a game of catand mouse, which we realized is
quite hard to win against alarge.
Even though we started firstand we did for one of our
clients.
They asked us to do acomparison, like a test, and we
actually found that ourautomated AI is actually better
(18:31):
than Google's one, which isquite fun, it's very impressive
and a bunch of benchmarks.
We actually found that ourautomated ai is actually better
than google's one, which isquite fun and that's very
impressive benchmarks.
We actually got higher accuracy.
And then we went head to headwith the schroeder's data fund
and beat them as well.
So they feel like, yeah, we'rebetter.
But the thing is, you can sayall that and have the proof, but
that isn't really the objectionthat people have.
It's like, well, no one gotfired for buying microsoft, you
know, or IBM, as they say.
(18:52):
So that just became a verychallenging thing.
So, yeah, we've kind of nicheddown a bit more into some of the
places.
We've had a lot more success.
So one would be like automotivefinancing.
So for brands like Hyundai andKia and Volvo, try to help them
understand which customers aregoing to renew, what car to sell
(19:12):
them and when we should talk tothem.
And that's proved to workreally well, and so there's a
lot of interesting depth to thatmarket that we're exploring now
and it's kind of a lot easierto sell something that's kind of
solving for a particularproblem, rather than having a
kind of platform that solveshorizontally for lots of
problems where you're saying,saying, yeah, we can help you do
(19:33):
anything.
You can build something to dolike with Deloitte we're
automating tax, charlotteTilbury were doing product
recommendation, did some stuffwith BT doing predictive
maintenance these are all.
They don't help you in the nextdeal really, because there's in
such a different market that itdoesn't really count for much.
So having momentum in a singlevertical is really what we've
(19:54):
realised is important.
Speaker 1 (19:55):
And you know, I know how hard it is to build a
startup.
Have you had moments whereyou've just gone this is too
hard, or a moment where you'vereally questioned, you know,
your decision to be in this typeof business.
Speaker 2 (20:08):
Yeah, absolutely.
I think the first three yearswere going from zero to
investment, and then you knowthe desperate scrambles to kind
of make that work before we runout of money.
Me and my co-founder wereworking every single day,
including weekends, not takingany holidays and working till
late in the evening for years,and that had a significant toll
(20:32):
on my mental well-being yeah,luckily not so much on his,
otherwise I think we'd betotally screwed.
That was incredibly miserableand I kind of realized at a
certain point I just needed toactually take time for myself to
do other things, like go to thegym, otherwise I was just
you're just not effective andyou've got a certain amount of
do things now like you justtypically will go to the gym.
Speaker 1 (20:54):
You're more intentional about managing
mental health exactly.
Speaker 2 (20:57):
Yeah, I think it's actually my number one priority,
like I have to fit the businessat number two because if I
don't, then I'm not effective atall.
You know that everything slipsaway and you're not motivated,
you're not happy, especially ifyou're doing anything customer
facing.
That's not a good place to be,even on the back end stuff.
If you hate your life, you'renot happy, especially if you're
doing anything customer facing.
That's not a good place to be,even on the backend stuff.
If you hate your life, you'renot effective at anything you
try and do.
(21:17):
So, yeah, I've realized yourmental health has to be the
number one priority and thatmeans, you know, having taken
some holiday, you know stoppingworking at a certain time of the
evening unless there'ssomething super important that
has to get over the line, andjust making sure there's a few
hours where I'm just able to notthink about business.
Yeah, and that's proved muchmore sustainable.
Yeah.
Speaker 1 (21:36):
I'm pleased you've raised that, actually, because I
was having that conversationearlier today about.
You know we've got inputs tobusiness that drive outputs but
actually some of those inputsare running at a sort of 0.5x in
terms of the leverage we getoff them if we're tired or
whatever, whereas others itmight be the effectiveness of
the hiring decisions we make orsomething are running at a 5x or
(21:58):
a 10x because we make gooddecisions and we hire good
people or we do things.
You know we've got a higherenergy level, so putting an
emphasis into your mental healthas the founder just feels
critical, like you know, reallyreally critical.
Has there been a decision thatyou've made over the last eight
years where you look back andyou think that was a really
(22:19):
pivotal decision and it was adecision that was really well
made and it's made quite animpact on the business?
Yeah, I mean what we just said.
Speaker 2 (22:26):
I mean prioritising mental health over anything else
probably would be the one.
For me, that's the number one,yeah, yeah, and I think from our
point of view as a businessmore abstractly, I think we
decided to stop going after bcmoney at a certain point and
start to just concentrate onrunning a business that's not a
(22:47):
tech, but when I say a techbusiness, I mean your typical
kind of silicon valley that'sraised loads of bc money and
hope we can keep raising andhope we get enough product,
enough customers, enough marketdominance, and then you know,
maybe we'll be billionaires ormaybe we'll be broke, and that
is not a good place to be,because you know VC's got 10
(23:07):
other companies in its portfolioand it doesn't care if you die,
it just wants one of them.
One of 10 need to win.
So they'll be pushing youtowards decisions that are not
in your own best of interest,and I know a bunch of other
founders that have been severelyscrewed through.
That you know.
In fact, I don't think I knowanyone that's taken bc money.
That's it's worked out for them.
I think that's on par with, ifyou look at the stats, it's the
(23:30):
you know the portfolioperformance for bc would be.
One does really well, a coupledo okay, and the rest are wiped
out.
And they're really wiped outbecause they'll take
preferential shares and thingslike that.
So you're not going to exitwith anything if you don't exit
with a valuation above X.
So as soon as we started tryingto run a business that was
(23:51):
self-sustaining and stood on itsown legs you don't have a
runway to worry about, you knowyou can make decisions in your
own time.
You're not trying to grow atall costs, at the cost of
profitability or good decisionmaking.
I think that was kind of bornoff the back of like trying to
understand the business.
I think ultimately, after thefirst four years and all the
(24:15):
stress and everything like that,I think me and my co-founder or
co-founders we had a co-founderjoin after the first year and
she's been amazing basically weall decided that really what we
want is to be able to at somepoint exit the business and, you
know, get some return, becauseit's so easy to not get anything
.
(24:36):
So many times it could haveresulted in a complete wipeout
and there's been so much painand blood and sweat put into it
now that it would be a tragedyif we weren't able to release
any of the equity.
So, instead of trying to buildsomething that's going to
compete with, necessarilylooking to something that's
going to be like a £100 billionvaluation, it'd be nice and
we're certainly not going to notdo that.
(24:56):
But the decisions you make thatlead towards that are often
incredibly risky and you couldmake a decision to go for some
fraction of that which is like10 times less risky.
You know, and that's not thedecision that bc would
necessarily like you to make.
But when you're like, okay,well, why don't we try and make
this business sustainable, growit and try and grow it in a way
(25:18):
that's not gonna push us intocrazy decisions that took a lot
of stress out of the wholeproceeding and allowed us to be
a lot more, you know,adventurous and you know we
we've started some interestingnew kind of like mini businesses
within the business, so likesomething called cortical chat
that we've built, which is basedon, you know, everyone.
(25:39):
As soon as we'd sort of seensome stuff around llms before,
chat gpt came out and we wereall super excited.
But as soon as chat gpt hit themainstream and had the reception
it did, we were like, well,obviously this is going to be a
massive thing.
Let's try and do somethingstraight away.
So we've got a kind ofsomething called cortical chat,
which is effectively itleverages a bunch of technology
(26:01):
on our platform but while ourplatform is like a sas
enterprise platform with areasonably high ticket price,
this is something where you canget it for 50 quid a month and
you can have a free trial andyou can just sign up with a
credit card and we don't need tobe involved in the sales
process.
So, instead of something whichis fundamentally limited by your
sales resource and doing a lotof outbound marketing to
(26:22):
generate and a lot of, you know,free consulting in some cases
to generate pipeline, this issomething where you can do some
ad spend.
People can come try it out andactually potentially sign up and
start paying you money withoutyou could be in bed while that
happens, which is kind of adream it's every founder's dream
.
Speaker 1 (26:42):
It's every founder's dream and so, interestingly so,
you had the two co-founders youand your other co-founder and
then you added a third one.
Did that change the dynamicsbetween you and your original
co-founder?
Did that make things differentmore?
Speaker 2 (26:55):
challenging easier Well, actually easier.
I think it was a sister who wason maternity leave from a sort
of marketing job at a big brandand she'd done a whole bunch of
helping out for basicallynothing.
And while me and my co-founderoriginal co-founder are not the
most, things like admin and HRand things like that are not our
(27:17):
forte necessarily.
We kind of tend to leave thingsto the last minute.
And you know, getting thingslike ISO 27K compliant and stuff
like that is not stuff that welike doing and we didn't know
anything about marketing andshe'd kind of got an ops
background as well, knowanything about marketing and
she'd kind of got an opsbackground as well.
So having someone that's kindof like almost like the core to
keeping the whole businessrunning and able to do those
(27:37):
kind of things yeah, she doesall the iso stuff, does a lot of
the sales back end like helpsus apply to grants, does there's
so many things that you knowyou should be doing but you're
involved in the tech or you'rejust underground doing sales
meetings but you might not beable to remember exactly what
follow-up to do or you're toobusy to do the follow-up
properly.
And having someone that wasable to do that and do it well,
(27:59):
was a massive relief for usreally.
So it was a kind of part of thepuzzle that wasn't really there
before, I think, so it wasgreat.
Speaker 1 (28:08):
And you've had 15 years in AI, and for most of us,
ai has just been sort of out ofthe landscape only for a few
years, and so you've had lots oftime in this space.
You've got a PhD, you know, indata analytics, and so you know
this stuff.
What sort of insight do youhave around where you see AI
going in the next couple ofyears?
Speaker 2 (28:29):
Perhaps that might be a bit controversial or
something that you, you know,you perceive, you know may play
out so if I didn't have most ofmy net worth in this business
and I was able to put bets onstocks, I would be looking for
ones investing in ai agents.
So what do I mean by that?
Let's think of something likesiri right or alexa.
(28:49):
Now you can say alexa, playthis song, but you can't say
alexa.
I'm thinking about getting aportugal in july.
I'd like to have villa with ahot tub.
Don't want to pay more thanthis needs to be within 50 miles
of a michigan style restaurantand give me, email me 10
potential options.
You know something like that.
(29:10):
It's not like each one of thosesteps isn't something that can
be done automatically.
It's that we don't really havea reasoning engine that can take
all of those requirements, turnthat into a plan of attack and
then solve for all of thosefairly kind of nebulous demands.
But really that is probablydoable with today's technology,
(29:33):
just about with things like LLMsand you know they're pretty
good at integrating into.
You can get an LLM to go on awebsite it's not seen before and
figure it out.
I think the next generation oflms or or maybe a new technology
of that ilk, which is, you knowto say, transformer based
neural networks, perhaps withmore of a kind of planning and
(29:54):
reasoning side to them, we'll beable to solve for things like
that.
And if you think about it,that's I've just given you an
example of being a travel agent,but there's so many other tasks
which you might need to do foryourself, as a civilian, let's
say.
But in business, how many tasksare reasonably simple but
actually could be fullyautomated with something like
(30:15):
that?
How many paralegals are justgoing around looking for various
things in various old cases,just going around looking for
various things in various oldcases?
Or you might have someonethat's organizing events for the
business or needs to get all ofthese various stakeholders into
the same room at the same time.
It's like when are you free?
Or when are you free, let'sfind somewhere, let's book a
place.
Here's the budget.
And I'm not saying this isnecessarily replacing any
(30:38):
particular job.
It's more like aspects ofcertain people's jobs probably
could be fully automated withthat and it would be kind of the
boring aspects.
Usually that's kind of.
I mean, an analogy might besomething like co-pilot for
coding.
Yes, yeah, that's quite.
It's not going to be designingsystems anytime soon or building
whole projects, but if you're Iuse it to kind of type complete
(31:00):
code and it's like, oh, thishas saved me like five minutes
here and five minutes there.
And you can ask it oh, whatwould you think the best way to
do this would be?
So you still need to havesomeone who knows what they're
doing controlling it.
But it's taking chunks out ofthe daily workload and you know
a small number here, a smallnumber there, and I think the
same will be.
That will be the same for kindof ai agents.
(31:22):
It will be.
Certain aspects of certainroles will be.
You can completely leave it tothe agent and it's like okay,
yep, these look reasonable,let's kill these two.
So, yeah, I think that's goingto be the next big thing.
And it will be a big thingbecause there'll be a kind of
consumer facing element to it,but probably the real value will
come from the boring backendadmin stuff that will be
(31:44):
automated as a result.
Speaker 1 (31:46):
And do you see those AI agents being heavily domain
specific?
So, for example, using theexample you gave, there would be
a travel one.
Yeah, there would be aparalegal one.
You know, is that how you seeit going, or is that a
limitation I'm putting on itthat actually doesn't need to be
there?
Speaker 2 (32:05):
so I mean we've been building a as part of cortical
chart framework for doing this.
Um again it's, you know we'reat the limits of the technology
as for the back end stuff likellms, but we've been doing stuff
across travel.
I mean, ultimately you need aframework where you can break
tasks down, an llm like likechat, gpt or GPT-4.
Rather, if you give it too muchfor any one task, it's never
(32:29):
going to do a good job, like themore instructions you give it,
the less task cohesion it has.
So it's it's like you've got tobe able to break the task down
and have agents pass bits alongto other agents, almost like a
network of various more specificthings.
So if you make an element of atask specific enough, then GPT-4
(32:49):
will do a pretty good job,pretty consistently.
So it's like how do you breakdown booking a flight or sorting
out a holiday into varioustasks?
Well, you know, the tasksthemselves might be different,
but the framework to arrangethose things can be a sort of
layer that works across a bunchof different domains.
Speaker 1 (33:07):
Right and sort of changing tack slightly.
I'm interested when we weretalking before about some of the
impact that venture capital hason founders and the inability
to control your strategy, reallybecause your capital strategy
has been outsourced to somebodyelse, essentially, what's the
scene like in the UK?
A lot of the listeners of thispodcast are in Australia and New
(33:27):
Zealand.
What's the scene like aroundfounders deciding to sort of
look at some more of a publicmarket approach in the early
stages?
So they've gone past the angels, they've got friends and family
, they've done angels, but theydon't want to go down the
venture path and they want tomake sure they get to control
their capital strategy alongsidetheir business strategy.
(33:48):
Is there much of an opportunityto get into public markets
early in the UK?
So do you mean?
Speaker 2 (33:52):
like listing on a stock exchange Listings.
Yeah, I'm not sure it's thateasy in the UK.
I mean, places like Canada havea stock market for very small
cap-sized companies, so we had alook about actually ipa in
there many years ago and, butnot, I'm not so much in the uk.
I don't think I'd also saythat's maybe not the best idea
if you're a smaller companybecause there's so much more
(34:13):
oversight needed.
Yeah, the regular, you've gotall the extra.
Yeah, you got your quarterlythings on display for everyone.
Speaker 1 (34:21):
Yeah, um, yeah and what about the UK startup scene
generally?
Like is the UK startup scene,you know, compared to where it
was two or three years ago?
Is it thriving, Is it sort ofbouncing back or is it still
struggling along a little bit?
Speaker 2 (34:34):
I think that it's been quite difficult.
Due to the interest rate andeverything like that.
I think VCs and other kind offinanciers have sort of pulled
back a lot of investment simplybecause, you know, it's the
economic conditions really.
So it's been a lot harder toraise money and companies that
were kind of relying on gettingsuccessive rounds have been kind
(34:57):
of forced to do down rounds andthings like that.
So that creates a bit of avicious circle where VCs are
kind of like less scared toinvest in the first place
because they don't want to havea down round.
So yeah, it's not been ideal.
I mean, as I say, we haven'treally been courting that for a
while, so I wouldn't say we havea figure massively on the pulse
, but that's kind of what I'veheard from others who have been
(35:18):
much more involved.
It's been pretty tough.
Speaker 1 (35:20):
Yeah, it's been pretty tough.
Yeah, and you you also talkedat the start of the show about
almost a philosophy shift thatwent along the lines of you know
, you worked out early onwhether it was pre university or
sort of, at the time you're atuniversity that by working hard
and by being focused, just theoutcomes just changed completely
.
What was it that shifted foryou?
(35:40):
Because that's a big change.
Most people don't make a shiftlike that.
They're either like that orthey're not.
What was it that changed?
Was it your father and passingaway, or what was involved there
?
Speaker 2 (35:51):
I think I was like what am I doing with my life?
Do you know what I mean?
I had all this opportunity andI'm kind of squandering it.
And I felt so muchself-loathing, if you like,
about that.
I was like I really don't wantto be this person anymore and I
would sort of write a plan ateach beginning of each week.
So I'm going to do this many.
I use Pomodoro.
So this many Pomodoro's of workon my business, this many on uni
(36:13):
, this many on other things likefitness or like boxing or
whatever, and the sense that ifI didn't achieve these goals I
was sort of sliding back intothat person I didn't want to be,
was enough of a kind of mentalcounterweight.
I think maybe if I hadn't hadsuch a profound negative
experience, that wouldn't havebeen there.
So it's like having getting tothat point, I really hated
(36:36):
myself for who I was in thesense of like I felt like I
wasn't achieving what I could beachieving, not in, not in a
more kind of general sense, butreally I was letting myself down
massively.
And as soon as I started doingthat, I started to notice a big
difference.
But funnily enough, you'dafford it, and so as soon as you
actually start working hardthings, things tend to improve
(36:58):
quite dramatically.
So I was like, getting fitter,I was doing better at uni, I was
getting my more consultingstuff off the ground, and then I
was like, well, if I ever stopdoing this, I'm gonna go back to
being that person.
So it was kind of like afeedback loop of the more I
progressed, the further I felt Igot from that person almost,
the more I could lose, thefurther I had to fall, the more
(37:18):
I was more committed to notdoing that, and so I could push
myself further because I waseven more committed to this new
path, if you see what I mean.
So that was the kind of mindsetI had.
Speaker 1 (37:29):
And do you ever slip back?
Is there ever a tendency toslip back to those old you know
almost childhood habits andthoughts, or is it actually so
hardwired in you now this change?
Speaker 2 (37:39):
I think, yeah, I've experimented with this and so I
find if I, you know, if I don'twrite down like a weekly set of
things, after a few weeksobviously, just, it's a lot
easier to stay motivated whenyou've got a team and a customer
and it's not, it's not really aquestion of self-motivation
anymore, because the externalworld motivates you, but other
(37:59):
stuff outside of work, gym andso forth, that that tends to
slack off.
And I think you do, habits dochange and I think you know I'm
a million miles different anyway, but improvement does stop, I
think.
So I found I'm just going tokeep doing it.
Yeah, and it's kind of like Ihave sort of a grand plan of
what I want to do in my life andeach, each week, if you like,
(38:22):
well, I'll do a little bittowards that various aspect of
it, this aspect, a sense ofbeing on plan and going towards
a purpose, and even if you'remaking small progress, that's
quite a powerful thing formental health and sense of
wellbeing.
So if you're, you know, at theend of each week you're like,
yes, I've got a bit closer to mygoals, I think that's pretty
powerful.
And like writing down a planand achieving it is on a
(38:44):
short-term basis is a great wayto give your brain that
nourishment of like yeah, I'm onthe right path, I am getting
closer to my goals.
Speaker 1 (38:51):
I mean, we naturally want to feel we've got a purpose
and I think a lot of us want tobuild something.
So by making steps to buildthat we can see we're building
something, that we can see we'rebuilding something, it's a
powerful motivator.
And what would you say for youand I hate the phrase zone of
genius, but let's use it what isyour superpower Like?
When Alex is working in thisspace, the business is going to
(39:15):
be a lot better off becauseyou're working in your
superpower.
Speaker 2 (39:17):
What is that?
Yeah, I mean for me it'dprobably be being kind of good
at the sales and the tech.
So I mean, for me it'd probablybe being kind of good at the
sales and the tech.
So I mean a lot of businessesstruggle where you've got a
sales team that don't understandthe capabilities.
So especially when you're doingsomething like where you're
right at the cutting edge, wherethe capabilities are not well
defined, so can this be solvedwith AI?
Speaker 1 (39:39):
or not.
Speaker 2 (39:39):
It's not an easy question to answer, but if
you're trying to sell a projectwhere, okay, we want you to
automate this aspect of ourbusiness, if you can be that
salesperson that alsounderstands the capabilities,
that's a pretty strong.
You can really.
The client can obviously tellthat you know what you're
talking about and you can answerany technical questions on the
fly and you cannot commit yourcompany or your team to this
(40:01):
completely unfeasible projectand, conversely, on the backend,
on the engineering side, youcan make decisions based on well
, is this going to get us tothese objectives that matter to
the business?
But you know, there's, I think,a tendency to overengine every
engineer.
I mean, I'd like to call myselfan engineer to some degree, to
(40:24):
some degree and the tendency isto just try and make everything
really awesome and robust andlike as in line with best
practice as possible, becausethat's you kind of want to do
what you've learned.
But that isn't necessarily thebest decision from a business
point of view.
You know, okay, if we, we don'tknow if this feature is going
to actually be worth building,so let's build a quick version
of it.
Okay, that might mean we haveto, like, take these shortcuts
in the short term or maybe notdo this thing.
(40:44):
We should do first, but wemight not need to do this at all
.
So sometimes developers,divorced of the business context
, can go down rabbit holes andover engineer the things that
you don't even need.
So kind of having a foot inboth camps, maybe the business
sales side and the tech side.
I think is is really powerfuland I think that's been probably
the probably the reason I'mhere, to be honest, because you
(41:06):
know you don't need to bebrilliant like overarching
genius to both either of thosethings, as long as you know
enough to be dangerous at both,at both yeah, and is that the
sort of thing that you wouldadvise a founder on?
Speaker 1 (41:17):
is you know you need to ideally be.
You don't have to be an expertat something that you it's good
if you're, if you're okay orgood at a couple of things like
what piece of advice have youbeen imparting to other?
You know younger founders orpeople starting out over the
last few years where you'velearned something and you think
you know I wish I'd known thisand so I tell lots of founders
this particular piece of advice.
(41:39):
What would that?
Speaker 2 (41:39):
be?
Yeah, exactly, I think.
I think it would probably betwo things.
Like one would be trying to askyourself, like why am I
starting this business?
What am I actually trying toget out of it?
Like yeah, actually, but beingsuper honest with yourself, is
it?
Are you trying to change theworld or do you want to make
money?
You know, do you want toprovide for your family and have
a comfortable life?
Because those take radicallydifferent paths, you know.
(42:01):
I think one would be the v money, one would be for God's sake,
don't get VC money becauseyou're gambling everything.
And the other would be to tryand if you're a hard skills
person when I say hard skills Imean a techie person let's say
try and put yourself out thereand be in more sales and
business situations and, if youcan like, do that in a way where
(42:21):
you're kind of it's just youand a client and you, you learn
those skills because that itwill help you appreciate.
You know, whatever camp you'rein, try and understand a bit
more about the other campthere's lots of.
I've got a friend who's 37 andhe's uh, you know, always been
in sales and he's he's doinglike a data science and python
course, and things.
That's just you know.
You can do these things forfree.
It doesn't take you very manyhours and you've gone from zero
(42:43):
to way more than zero and it'squite a good investment in time
and energy and just makes youway better at appreciating what
you maybe you're asking of otherpeople when you're making a
sale or, you know, creating arequirement and things like that
.
So, yeah, that would probablybe my, my advice.
Speaker 1 (43:00):
Yeah and how do you inspire yourself?
You, I think I saw somethingabout you do some boxing.
Is that right?
You do boxing at the gym, ordid.
Speaker 2 (43:10):
Well, yeah, it's a bit of a story there.
So at school I was always a bitof a nerd, so I never really
stuck up for myself and it wasalways a part of me.
I didn't really like, I sort ofhated it, in fact, as you
probably could imagine.
And I was out in um on holidaywith my mates in Croatia and one
thing led to another and I gotstarted on by some fella in a
bar and I didn't stand up formyself at all and I think as a
(43:32):
result of that we ended upgetting piled on by like 30 guys
and I got quite badly hurt andI was like if I'd have just not
backed down in that situation,I'm fairly sure that wouldn't
have happened got a sort ofsmelt weakness, if you like.
So, like what can?
What's the single most hardcoreway I could make myself tough?
It was like right, I'm justgonna join a boxing gym and it
was incredibly tough.
(43:53):
You know, I was in the firstcouple of sparring sessions.
I was like afraid to like hit,even hit people because I just
felt so wrong.
But after a year or so, doingthat I was, I was completely
cured of that part of me andthat was actually really nice, a
really nice feeling ofsomething that's always you've
hated about yourself, kind ofsolving that and just feeling
like I'm actually not thatperson.
I could never go back to beingthat person.
(44:14):
That would not stand up forthemselves.
So that was awesome.
And I did boxing for a bunch ofyears and my dad had
Alzheimer's and both of hissisters had it and I'm like
maybe I shouldn't be getting hitin the head so much.
So I was like maybe I'll stopthat.
So currently I'm just going tothe gym a lot.
I've got ambitions to get intoBJJ and some of the wrestling
(44:34):
side of mixed martial arts youcould take on Mark Zuckerberg,
yeah exactly.
Yeah, I think I've got about 20kilos on Zach.
It'll be fine.
Speaker 1 (44:46):
Yeah, I think most humans are heavier than Mark
Zuckerberg.
You could be some sort of bigAI challenge.
That would be quite funny.
And do you find that?
It's just a complete antidoteto startup life and if you've
had a bad day, you can smack thecrap out of a bag, or you know?
I mean because you're notsparring now, presumably because
you don't want to get smackedin the head.
Speaker 2 (45:06):
Yeah, I mean, it certainly is.
I mean any kind of I think anykind of goal outside of startup
life is a great thing to haveand it doesn't need to.
Obviously it doesn't eversupersede your business, but,
like my co-founder, for example,started doing tennis, he's now
got into the advanced league andhe didn't start until with you
know, he's about eight yearsolder than me and he didn't even
start till we're like fouryears into the business and he's
(45:28):
super competitive and it's likesomething where he can you know
, progress and it's just a timeto switch off.
You know, for me at the momentit's just like I'm trying to.
I always wanted to get into likegreat shape at the gym and then
I sort of for some reason,right, I was doing this really
seriously for like a year beforefounding the business and then
my thought was like, well, we'llsell the business in like two
years and I'll be like on ayacht and I'll have like a
personal trainer.
(45:49):
It's like, obviously doesn'twork out like that.
But it was like seven years ofdoing absolutely no exercise at
all.
And then I was like, right, youknow what, I'm going to start
working towards this, eventhough I'm still doing a
business and it's like found.
You know, I'm going to the gymabout six times a week and you
know, doing all the proteinburners and so forth.
So just trying to get into goodshape has been my main goal.
(46:09):
I mean the boxing side ofthings.
I'm trying to help my cousin atthe moment train for a fight
and that's quite fun.
But, yes, I think anythingwhere you can see improvements
in yourself or some other aspectof your life that can give you
a, I think, fundamentally humanpsychology, if you see growth
somewhere, that's an incrediblynourishing thing and it's an
incredibly powerful way ofprotecting your mental health.
(46:31):
So, yeah, if you can be doinganything where you want to be
doing, it ideally a goal as well.
If yours and you're makingsmall progress towards it, then
that can be like a lifeline ofeverything's getting
overwhelming in the office.
That towards it, then that canbe like a lifeline of
everything's gettingoverwhelming in the office.
Speaker 1 (46:43):
That is a really wonderful way to finish and
that's superb advice for peoplelistening and you know Mark
Zuckerberg, if you are listening, you know Alex is ready for you
.
He is ready to fight you.
He is ready to go.
Bring it on, Bring this on.
So, Alex, has been absolutelyfantastic chatting to you today.
Thank you so much for the timeyou've given.
What I'll do is I'll make surethat I put some connections to
(47:06):
you and to Cortical in the shownotes so that if people are
interested in learning moreabout Cortical or wanting to
reach out to you personally,they can do so in the show notes
and connect with you.
Again, thank you so much foryour time.
It's been a real privilege anda real honour chatting to you
today.
So, yeah, thank you.
Thanks so much for having me.
What piece of advice have you been imparting to other
younger founders or peoplestarting out over the last few
years where you've learnedsomething and you think you know
?
I wish I'd known this.
Speaker 2 (00:12):
One would be trying to ask yourself like why am I
starting this business?
What am I actually trying toget out of it?
Like, are you trying to changethe world or do you want to make
money?
Do you want to provide for yourfamily and have a comfortable
life?
Because those take radicallydifferent paths.
Speaker 1 (00:27):
Alex is a chief data scientist and the co-founder of
Cortical.
Cortical is the automatedmachine learning platform for
professional data scientists.
Speaker 2 (00:38):
I think, anything where you can see improvements
in yourself or some other aspectof your life that can give you
a, I think, fundamentally humanpsychology.
If you see growth somewhere,that's an incredibly nourishing
thing, a sense of being on planand going towards a purpose, and
even if you're making smallprogress, that's quite a
powerful thing for mental healthand sense of well-being.
(00:59):
So if you're, you know, at theend of each week you're like,
yes, I've got a bit closer to mygoals.
Speaker 1 (01:05):
That's superb advice for people listening.
Hey everybody, it's GregSheehan.
Welcome to my podcast, whereyou will hear from a range of
guests, including those from thestartup world and those that
have had incredibly interestinglives and some stories to tell.
I would really appreciate it ifyou could hit the follow button
and share this amongst yourfriends, but, as you know, time
(01:27):
is limited, so let's get on withit and hear from our next guest
.
My guest today is Dr Alex Allen.
Alex is a chief data scientistand the co-founder of Cortical,
and we will get into that as theshow goes on.
Cortical is the automatedmachine learning platform for
professional data scientists.
Welcome to the show, alex.
Great to be here.
(01:47):
Thanks for having me.
It's a real pleasure to haveyou here.
We were introduced by somebodythat believes very strongly in
you and said look, I needed tointerview you.
In fact, she was quite adamantthat I needed to interview you.
She just absolutely loves whatyou do and very impressed by you
.
I generally start these bytalking a little bit about your
origin before we start to getinto Cortical and the story
(02:08):
around that, but I'd love toknow what sort of early
upbringing you had.
Were you always that kind ofentrepreneurial type?
Were you always a math-y kid?
What's the origin story?
Speaker 2 (02:17):
for you.
Yeah, that's a good one.
I mean I think I went to.
I grew up in Essex, which isjust sort of east of London, so
it's about a 50-minute drivefrom London, but it was
relatively rural.
I went to a comprehensive schoolwhich is like a state school,
and I was like it's kind of likefairly academic, but one of
those people that never reallyworked hard enough, if you know
what I mean.
I do, and I did things likecomputer science and business at
(02:40):
college.
And then I was looking fordegrees you could do and it was
like one degree in the country.
It was called cybernetics andartificial intelligence.
I just thought that soundsincredibly badass.
I'm just gonna have to do that,as I think that was.
This is in 2006, so you knowthis is way before AI was really
kind of a big deal.
But I was always a big fan ofscience fiction and I've always
(03:02):
thought that you know AI isgoing to be like it is in the,
in the movies.
You know when you're talking toa robot or something like from
Star Trek.
So I was always keen to kind ofget in on it and at first at
uni I was.
I struggled a little bitbecause I think there's a level
of oversight you get at collegeand things which you just at uni
you can just like you don'thave to go to lectures.
So I ended up kind of goingthrough some personal events
(03:24):
like my father passed away andthings like that, which kind of
led me to take a few monthsbreak.
And then I got into kind ofreading like self-help books and
they range from incrediblytrite to sometimes quite
profound, and through all ofthat I kind of came up with a
kind of personal philosophyaround setting myself goals and
targets and holding myselfaccountable.
(03:46):
And then I ended up just beingsomeone that worked really hard,
which is kind of I don't knowif it sounds realistic, but it
turns out you can changeyourself from someone that's you
know, never that I could, youknow self-control or
self-discipline into someonethat does have it.
And I found, when I was able tokind of consistently apply
myself and I think consistencyis like the really key thing
that I was doing a lot better.
(04:06):
And my third year at uni that Iwon a scholarship to do a PhD in
what was called data miningback then, but you'd probably
call it data science or AI now.
It's probably the same thingand that went well.
And, yeah, I was kind of duringduring the course of my life
I've always been obsessed withtrying to start a business and
did various things like fromselling memory phone mattresses
on e-commerce solutions to, youknow, building websites for
(04:28):
people and things like that.
So, kind of towards the end ofmy PhD, I just started doing
consulting in London.
I just said, yeah, I'm an AIconsultant.
And then, you know, because noone was doing it and and
actually I did legitimately haveexperience in a field that you
know this is this is back in2014, 15, there was really no
one else doing it and I was ableto do some cool stuff with some
(04:49):
interesting companies, whichgave me like a reasonably
interesting portfolio.
By the time I'd sort ofgraduated uni sorry, my PhD,
really and then I was obsessedwith trying to start a tech
business and I ended up bumpinginto a chap was giving a giving
a talk at a conference and therewas this fellow at the back
drinking, drinking the free wine, and I was like he's probably
(05:10):
my guy.
So me and him got chattingduring the rest of the talks and
it turns out he was like anex-McKinsey.
He did sort of systemsoptimization, but he did it just
for himself and we got onreally well and he was a lot
older than me, like 25 yearsolder, but we we were like,
right, cool, let's start abusiness together.
And then we kind of there was afew other people in my circle
and then we, we all got togetherand started this, this business
(05:30):
which was around.
It sounds it was kind of like aboring application domain, but
the technology was quite cool.
We sort of took two days out andsort of brainstormed what we
could do and we came up with theidea of but, like call centers,
right, you have to go through alist of leads and and call them
and they just randomly assignthem to agents.
Um as well, what if we couldlearn about what an agent might
(05:53):
be good at in terms ofdemographics?
So let's say you're you know,I've got a fixed scottish accent
you might actually resonatebetter with people at the north
of the country, for example, andthings like time of day, like
if you're calling someone whoworks in the city at like three
in the afternoon they'reprobably not going to be in, but
if you're calling someone thatlives in the suburbs, there's a
chance, there's a you know, anew parent or someone that's
(06:13):
going to actually answer thephone.
So we were like, okay, we couldoptimize the time of day and
the right agent, then we cancreate something that could
basically, just throughallocating these leads
differently, create an an upliftin sales, and we started
working on the technology.
Really.
It was at that point I kind ofbumped into really my current
co-founder and that wascompletely random.
(06:35):
It was like our mutualex-partners invited us both to
this bonfire night.
They weren't ex-partners at thetime, so it wasn't something
weird like that and we juststarted talking about neural
networks and ai stuff.
And he was a.
He was working at barclaysdoing big distributed kind of
machine learning things, youknow big investment banks.
So, okay, cool, this guy'slegit um, and I've eventually
(06:56):
invited him to join this, thisnow quite large group of
founders there's like five of usand then we kind of first we as
we were building the technologyand we were going to try.
You know we got a few trialswith a few call centers and it
went really well.
But we kind of found that themarket was like a terrible one
because these, these operationsoften run on the knife edge of
profitability, so they're notreally keen on dropping 50 grand
(07:19):
a year on some untested machinelearning solution which no one
believed in.
No one knew what ai was, no oneknows what machine learning was
.
You'd have to start every pitchwith like a 20 minute sort of
educational session on.
I know machine learning isgoing to be really big, it will
really help you.
And these are kind of like hardbits and sales people.
They're like I don't like this,I don't know what this is, what
are you even trying to sell me?
(07:39):
We're like we'll just do it forfree then.
Yeah, but yeah, but you know,to do it for free we'll need to
integrate into your coolingsystem.
And it wasn't really like a niceeasy path to demonstrating the
technology.
And so we were trying to raisesome money to looking to VCs and
angels and a bunch of them weresaying like well, the kind of
technology we built to build themachine learning itself, we'd
(08:00):
kind of ended up.
We ended up realizing thatevery single time we had to do
one of these POCs, we had tokind of build a totally new
machine learning model fromscratch, because the product
will be different, thedemographics will be different,
might be similar sort of datainputs, but realistically you're
going to need maybe a totallydifferent type of model and
certainly a different set ofmodel parameters.
So we started building a kindof automated machine learning
(08:22):
framework to kind of speed thesethings up.
And that got to a point whereit was a kind of automated
machine learning framework tokind of speed these things up.
And that got to a point whereit was actually kind of fairly
advanced and these vcs were likewhy are you not just selling
that?
Why are you trying to sell thecall centers?
Um, this was before somethingcame into force called gdpr,
which is the uk's kind of datalegislation.
I think it was originally aeuropean thing and now it's just
(08:43):
the uk's, but I think it's asimilar thing to happen in
Europe, and that basically meantyou couldn't cold call people.
I mean, you still get loads ofcold calls now, but it's
technically illegal.
So isn't what you'd call anideal market, one that's
effectively being legislated outof existence?
And so me and my currentco-founder decided to kind of
leave this.
We kind of brought everyone out.
(09:04):
It was kind of pre-revenue, soit wasn't a huge the company, it
wasn't like we needed to investmillions in that and we started
doing basically taking intechnology, because we brought
the other guys out, we had thetechnology.
I mean me and him had built thewhole thing more or less anyway
, and this was 2016.
And then I mean we had a fewcontacts and a few people we'd
(09:25):
met.
And again, this was back when,if you Googled AI consultancy
London, we were the top resultwith basically no marketing
spend, just because we were theonly people saying we did it.
And we ended up landing ourfirst job, which was working
with another consultancy akin tosomeone like Accenture it
wasn't them um, to basicallybuild the AI roadmap for one of
(09:50):
the UK's largest banks.
So it was like a huge projectto build this whole massive AI
roadmap.
It was basically our first realjob, which is, you know to, by
today's standards, are likeutterly insane.
Um, you know really hard andyou know put a lot of hours in
and ultimately it was consideredreasonably successful and then,
you know, everything kind ofsprang off the back of that.
(10:11):
So, yeah, that was kind of theorigin story.
Really, I kind of veered moreinto the company there, but it's
the two, that kind of I guess.
Speaker 1 (10:18):
I'm so used to giving this talk.
Yeah, no, it's perfect.
And so you get started and youwin a bank as a customer.
I mean, how would you describewhat Cortical is now Like if you
were standing there and you had30 seconds to pitch what
Cortical does?
What does it do?
Speaker 2 (10:35):
We reduce the cost, risk and time of going from a
data and an idea into somethingthat lives in production in the
real world.
That's making you money usingmachine learning.
That would be the very, veryelevated pitch.
I mean, what do we actually do?
Well, we've got a platform thatwe've built over.
The same kind of engine that Iwas talking about from the
previous company is still thecore of our product and we have
(10:58):
a kind of mix of clients likecharlotte tilbury fashion brand.
They use us as really aplatform that they build things
on themselves.
Other companies such asDeloitte.
We've been heavily invested inhelping them, working with them
to kind of build the product aswell.
It's still underlined by ourtechnology.
So we kind of got a bit of amix of self-serve and more
(11:19):
consultancy slash kind of pointsolution led customers.
But ultimately, yeah, we buildeverything on our own technology
and as a result, we do things alot faster and we can do POCs
for a lot quicker.
And you know we can often dosome of the initial work
basically for free because it'sso quick for us to do.
So you know, a normal datascience project, like if you
(11:41):
were to come to me and say, ohokay.
So I want you to try andpredict the views for my podcast
.
You know, going forward,depending on who you go to, that
could be like a three-month bitof work to kind of come up, you
know, with an exploration phaseof data, a kind of data
analysis phase, initial modelbuilding, and these things are
all done manually and they allrequire kind of iteration and
(12:03):
like experimental feedback loops, and we can because our
platform kind of closes a loopon a lot of these things we can
kind of do that very, veryquickly.
So we could probably, in like aday or two after getting the
data, have some idea of howpossible that is and what kind
of accuracy we might get.
So we've been, as a result, beenable to kind of take projects
where I mean the normal successrate for ai projects, isn't it,
(12:24):
and by the success I mean takingsomething from someone's idea
and actually turning that intoproduction.
It's like less than 10according to gartner.
We're in like the 90 region ofsuccess, and that's really
because we take only take thingson.
We think we're going to be ableto do.
We are able to, because of ourplatform, do a bunch of the
initial kind of sniffing out of.
Is this viable for as part ofthe pre-sales process?
(12:47):
So when we get to that kind ofyou know actual project, we've
got confidence that we're ableto deliver it.
Speaker 1 (12:53):
And is it a mix of a services business and a software
business?
Do you do consulting around theoutside of this as well?
Speaker 2 (13:01):
Yeah, the mix of consulting to just SaaS has been
veering in the direction ofSaaS, which is good because
that's more scalable, but Ithink ultimately we're still in
a market where most companiesaren't doing really anything in
production with AI.
Some companies are A lot havestarted to think about it, but a
lot are still unsure as towhere to get started.
(13:23):
So consultancy is a part ofhelping companies through that
process and we've got eightyears of experience of doing
that now, which is quite a lotin such a new field.
So we do have a lot of value tooffer there.
And when we started thisbusiness, when we first came up
with this idea for the platform,there was annoyingly although
we didn't know about it at thetime one American company called
(13:44):
Data Robot that was alreadydoing it.
This is always one that getsthere first and you know, we
completely independently came upwith the same idea and over the
course of lifespan, google,amazon and Microsoft have all
released equivalents to whatwe've got, which makes the sales
process more challenging.
Right, because these companieshave already got big,
longstanding relationships withthese large, trusted providers
(14:08):
and they've got all through thedata security and it's like oh,
you want to enable this elementof Google Vertex, for example.
You want to enable that.
That's just.
You've already got that throughyour data security, you've
already got it through your ITops and whatever.
You just tick a few boxes andit's there, whereas we've got to
say, look, no, we'll go throughall that process.
So us bringing that personalelement.
(14:29):
So if you go within those bigplayers bringing that personal
element, so you know, if you gowith any of those big players,
they're not going to give youanywhere near the level of
handholding and help that you'llget.
And if they are, if they, ifyou do pay for the consulting
package, it's not going to bethe founder or anyone close to
that or be, you know, relativelyjunior person who's going to
have a lot less experience thanus.
So that's, it's become a pathfor us to differentiate
(14:50):
ourselves and to provide a kindof better quality of service.
But we are seeing as morecustomers just realize they need
things like what we're doing,that we have to do less of that,
which is good, because it'sharder to scale a consultancy,
obviously, and keep the qualityhigh.
Speaker 1 (15:06):
It's interesting because competing with the big
players now in AI must be justalmost impossible to take them
on at their own game.
So to personalise it and, asyou say, be the founder who's
potentially leading some of thestrategic conversation, is the
point of difference for you.
So just give us a bit of asense of what does the company
(15:27):
look like now, to the extentthat you're willing to share.
So you started with the two ofyou, and then what are we?
Eight years on?
What is the company?
Sort of shape does it have now?
Speaker 2 (15:35):
Yeah, so after two years of starting the business
we raised $1.2 million.
This was from angels and thenwe scaled the team to like 15,
16 people and I think me and myco-founder are both really good
at building their technology andbuilding DOCs and doing their
sales, which is kind of a nicemix as a starting team.
(15:57):
But one of the things, whilewe're good in front of people
and we're good at selling theproduct, that doesn't
necessarily translate to beingable to scale a sales team and
find a product, understandproduct market fit and there's a
whole bunch of stuff that, as afounder yourself, you probably
know about the journey there.
Right, it's not easy and whenyou get a huge wad of investment
(16:19):
you've got some chance to testand learn and make some mistakes
.
But you know, ultimately thereason most businesses fail is
because there's a learning curveand a runway and you've got to
sort of get to that point in thelearning curve while you still
have some money left and wemanaged to do that.
But we also had a huge, Iwouldn't say catastrophe.
(16:40):
But you know, when COVID hit wehad about 50% of our revenue.
That was, you know, we prettymuch closed the deal but not
signed a contract for a largeset of really big projects and
then, as soon as covid hit,every single business was like,
yeah, no, innovation budget,we're gonna hunker down.
So we ended up with a big holein our cash flow so we had to
(17:03):
scale the team back a little bit.
Um, so we're now about 10people, but we've kind of got to
the point where we don't have arunway, we've self-sustaining
on our own revenue, which is amuch, much less stressful place
to be, and I'd probablyrecommend getting to that place
as soon as possible as abusiness owner, because you're
just then able to do moreexperimentation.
And we've now got to a placewhere we're much closer, closer
(17:26):
and or have got some realevidence, I think, of product
market fit, which is somethingwe, you know, we still, we still
closing deals, but it was kindof all quite eclectic and they
weren't, you know, it was more,maybe, a product of the nascence
of the market, rather than wefound something solidly we could
repeatedly win at.
And I think, you know, thefrustrating thing for us was we
(17:47):
were always like three stepsahead in terms of what we wanted
to build, versus someone likeMicrosoft, and we knew exactly
what they were going to buildbecause we were thinking well,
this is obviously the next stepand they would have a team of
like 50 or 100 people buildingtowards it and we could spare
like one developer.
So well, we had the ideas first, the execution often we can get
a basic version out, but theneverything they release would be
(18:10):
already integrated into a wholesuite of other products.
So it just became a game of catand mouse, which we realized is
quite hard to win against alarge.
Even though we started firstand we did for one of our
clients.
They asked us to do acomparison, like a test, and we
actually found that ourautomated AI is actually better
(18:31):
than Google's one, which isquite fun, it's very impressive
and a bunch of benchmarks.
We actually found that ourautomated ai is actually better
than google's one, which isquite fun and that's very
impressive benchmarks.
We actually got higher accuracy.
And then we went head to headwith the schroeder's data fund
and beat them as well.
So they feel like, yeah, we'rebetter.
But the thing is, you can sayall that and have the proof, but
that isn't really the objectionthat people have.
It's like, well, no one gotfired for buying microsoft, you
know, or IBM, as they say.
(18:52):
So that just became a verychallenging thing.
So, yeah, we've kind of nicheddown a bit more into some of the
places.
We've had a lot more success.
So one would be like automotivefinancing.
So for brands like Hyundai andKia and Volvo, try to help them
understand which customers aregoing to renew, what car to sell
(19:12):
them and when we should talk tothem.
And that's proved to workreally well, and so there's a
lot of interesting depth to thatmarket that we're exploring now
and it's kind of a lot easierto sell something that's kind of
solving for a particularproblem, rather than having a
kind of platform that solveshorizontally for lots of
problems where you're saying,saying, yeah, we can help you do
(19:33):
anything.
You can build something to dolike with Deloitte we're
automating tax, charlotteTilbury were doing product
recommendation, did some stuffwith BT doing predictive
maintenance these are all.
They don't help you in the nextdeal really, because there's in
such a different market that itdoesn't really count for much.
So having momentum in a singlevertical is really what we've
(19:54):
realised is important.
Speaker 1 (19:55):
And you know, I know how hard it is to build a
startup.
Have you had moments whereyou've just gone this is too
hard, or a moment where you'vereally questioned, you know,
your decision to be in this typeof business.
Speaker 2 (20:08):
Yeah, absolutely.
I think the first three yearswere going from zero to
investment, and then you knowthe desperate scrambles to kind
of make that work before we runout of money.
Me and my co-founder wereworking every single day,
including weekends, not takingany holidays and working till
late in the evening for years,and that had a significant toll
(20:32):
on my mental well-being yeah,luckily not so much on his,
otherwise I think we'd betotally screwed.
That was incredibly miserableand I kind of realized at a
certain point I just needed toactually take time for myself to
do other things, like go to thegym, otherwise I was just
you're just not effective andyou've got a certain amount of
do things now like you justtypically will go to the gym.
Speaker 1 (20:54):
You're more intentional about managing
mental health exactly.
Speaker 2 (20:57):
Yeah, I think it's actually my number one priority,
like I have to fit the businessat number two because if I
don't, then I'm not effective atall.
You know that everything slipsaway and you're not motivated,
you're not happy, especially ifyou're doing anything customer
facing.
That's not a good place to be,even on the back end stuff.
If you hate your life, you'renot happy, especially if you're
doing anything customer facing.
That's not a good place to be,even on the backend stuff.
If you hate your life, you'renot effective at anything you
try and do.
(21:17):
So, yeah, I've realized yourmental health has to be the
number one priority and thatmeans, you know, having taken
some holiday, you know stoppingworking at a certain time of the
evening unless there'ssomething super important that
has to get over the line, andjust making sure there's a few
hours where I'm just able to notthink about business.
Yeah, and that's proved muchmore sustainable.
Yeah.
Speaker 1 (21:36):
I'm pleased you've raised that, actually, because I
was having that conversationearlier today about.
You know we've got inputs tobusiness that drive outputs but
actually some of those inputsare running at a sort of 0.5x in
terms of the leverage we getoff them if we're tired or
whatever, whereas others itmight be the effectiveness of
the hiring decisions we make orsomething are running at a 5x or
(21:58):
a 10x because we make gooddecisions and we hire good
people or we do things.
You know we've got a higherenergy level, so putting an
emphasis into your mental healthas the founder just feels
critical, like you know, reallyreally critical.
Has there been a decision thatyou've made over the last eight
years where you look back andyou think that was a really
(22:19):
pivotal decision and it was adecision that was really well
made and it's made quite animpact on the business?
Yeah, I mean what we just said.
Speaker 2 (22:26):
I mean prioritising mental health over anything else
probably would be the one.
For me, that's the number one,yeah, yeah, and I think from our
point of view as a businessmore abstractly, I think we
decided to stop going after bcmoney at a certain point and
start to just concentrate onrunning a business that's not a
(22:47):
tech, but when I say a techbusiness, I mean your typical
kind of silicon valley that'sraised loads of bc money and
hope we can keep raising andhope we get enough product,
enough customers, enough marketdominance, and then you know,
maybe we'll be billionaires ormaybe we'll be broke, and that
is not a good place to be,because you know VC's got 10
(23:07):
other companies in its portfolioand it doesn't care if you die,
it just wants one of them.
One of 10 need to win.
So they'll be pushing youtowards decisions that are not
in your own best of interest,and I know a bunch of other
founders that have been severelyscrewed through.
That you know.
In fact, I don't think I knowanyone that's taken bc money.
That's it's worked out for them.
I think that's on par with, ifyou look at the stats, it's the
(23:30):
you know the portfolioperformance for bc would be.
One does really well, a coupledo okay, and the rest are wiped
out.
And they're really wiped outbecause they'll take
preferential shares and thingslike that.
So you're not going to exitwith anything if you don't exit
with a valuation above X.
So as soon as we started tryingto run a business that was
(23:51):
self-sustaining and stood on itsown legs you don't have a
runway to worry about, you knowyou can make decisions in your
own time.
You're not trying to grow atall costs, at the cost of
profitability or good decisionmaking.
I think that was kind of bornoff the back of like trying to
understand the business.
I think ultimately, after thefirst four years and all the
(24:15):
stress and everything like that,I think me and my co-founder or
co-founders we had a co-founderjoin after the first year and
she's been amazing basically weall decided that really what we
want is to be able to at somepoint exit the business and, you
know, get some return, becauseit's so easy to not get anything
.
(24:36):
So many times it could haveresulted in a complete wipeout
and there's been so much painand blood and sweat put into it
now that it would be a tragedyif we weren't able to release
any of the equity.
So, instead of trying to buildsomething that's going to
compete with, necessarilylooking to something that's
going to be like a £100 billionvaluation, it'd be nice and
we're certainly not going to notdo that.
(24:56):
But the decisions you make thatlead towards that are often
incredibly risky and you couldmake a decision to go for some
fraction of that which is like10 times less risky.
You know, and that's not thedecision that bc would
necessarily like you to make.
But when you're like, okay,well, why don't we try and make
this business sustainable, growit and try and grow it in a way
(25:18):
that's not gonna push us intocrazy decisions that took a lot
of stress out of the wholeproceeding and allowed us to be
a lot more, you know,adventurous and you know we
we've started some interestingnew kind of like mini businesses
within the business, so likesomething called cortical chat
that we've built, which is basedon, you know, everyone.
(25:39):
As soon as we'd sort of seensome stuff around llms before,
chat gpt came out and we wereall super excited.
But as soon as chat gpt hit themainstream and had the reception
it did, we were like, well,obviously this is going to be a
massive thing.
Let's try and do somethingstraight away.
So we've got a kind ofsomething called cortical chat,
which is effectively itleverages a bunch of technology
(26:01):
on our platform but while ourplatform is like a sas
enterprise platform with areasonably high ticket price,
this is something where you canget it for 50 quid a month and
you can have a free trial andyou can just sign up with a
credit card and we don't need tobe involved in the sales
process.
So, instead of something whichis fundamentally limited by your
sales resource and doing a lotof outbound marketing to
(26:22):
generate and a lot of, you know,free consulting in some cases
to generate pipeline, this issomething where you can do some
ad spend.
People can come try it out andactually potentially sign up and
start paying you money withoutyou could be in bed while that
happens, which is kind of adream it's every founder's dream
.
Speaker 1 (26:42):
It's every founder's dream and so, interestingly so,
you had the two co-founders youand your other co-founder and
then you added a third one.
Did that change the dynamicsbetween you and your original
co-founder?
Did that make things differentmore?
Speaker 2 (26:55):
challenging easier Well, actually easier.
I think it was a sister who wason maternity leave from a sort
of marketing job at a big brandand she'd done a whole bunch of
helping out for basicallynothing.
And while me and my co-founderoriginal co-founder are not the
most, things like admin and HRand things like that are not our
(27:17):
forte necessarily.
We kind of tend to leave thingsto the last minute.
And you know, getting thingslike ISO 27K compliant and stuff
like that is not stuff that welike doing and we didn't know
anything about marketing andshe'd kind of got an ops
background as well, knowanything about marketing and
she'd kind of got an opsbackground as well.
So having someone that's kindof like almost like the core to
keeping the whole businessrunning and able to do those
(27:37):
kind of things yeah, she doesall the iso stuff, does a lot of
the sales back end like helpsus apply to grants, does there's
so many things that you knowyou should be doing but you're
involved in the tech or you'rejust underground doing sales
meetings but you might not beable to remember exactly what
follow-up to do or you're toobusy to do the follow-up
properly.
And having someone that wasable to do that and do it well,
(27:59):
was a massive relief for usreally.
So it was a kind of part of thepuzzle that wasn't really there
before, I think, so it wasgreat.
Speaker 1 (28:08):
And you've had 15 years in AI, and for most of us,
ai has just been sort of out ofthe landscape only for a few
years, and so you've had lots oftime in this space.
You've got a PhD, you know, indata analytics, and so you know
this stuff.
What sort of insight do youhave around where you see AI
going in the next couple ofyears?
Speaker 2 (28:29):
Perhaps that might be a bit controversial or
something that you, you know,you perceive, you know may play
out so if I didn't have most ofmy net worth in this business
and I was able to put bets onstocks, I would be looking for
ones investing in ai agents.
So what do I mean by that?
Let's think of something likesiri right or alexa.
(28:49):
Now you can say alexa, playthis song, but you can't say
alexa.
I'm thinking about getting aportugal in july.
I'd like to have villa with ahot tub.
Don't want to pay more thanthis needs to be within 50 miles
of a michigan style restaurantand give me, email me 10
potential options.
You know something like that.
(29:10):
It's not like each one of thosesteps isn't something that can
be done automatically.
It's that we don't really havea reasoning engine that can take
all of those requirements, turnthat into a plan of attack and
then solve for all of thosefairly kind of nebulous demands.
But really that is probablydoable with today's technology,
(29:33):
just about with things like LLMsand you know they're pretty
good at integrating into.
You can get an LLM to go on awebsite it's not seen before and
figure it out.
I think the next generation oflms or or maybe a new technology
of that ilk, which is, you knowto say, transformer based
neural networks, perhaps withmore of a kind of planning and
(29:54):
reasoning side to them, we'll beable to solve for things like
that.
And if you think about it,that's I've just given you an
example of being a travel agent,but there's so many other tasks
which you might need to do foryourself, as a civilian, let's
say.
But in business, how many tasksare reasonably simple but
actually could be fullyautomated with something like
(30:15):
that?
How many paralegals are justgoing around looking for various
things in various old cases,just going around looking for
various things in various oldcases?
Or you might have someonethat's organizing events for the
business or needs to get all ofthese various stakeholders into
the same room at the same time.
It's like when are you free?
Or when are you free, let'sfind somewhere, let's book a
place.
Here's the budget.
And I'm not saying this isnecessarily replacing any
(30:38):
particular job.
It's more like aspects ofcertain people's jobs probably
could be fully automated withthat and it would be kind of the
boring aspects.
Usually that's kind of.
I mean, an analogy might besomething like co-pilot for
coding.
Yes, yeah, that's quite.
It's not going to be designingsystems anytime soon or building
whole projects, but if you're Iuse it to kind of type complete
(31:00):
code and it's like, oh, thishas saved me like five minutes
here and five minutes there.
And you can ask it oh, whatwould you think the best way to
do this would be?
So you still need to havesomeone who knows what they're
doing controlling it.
But it's taking chunks out ofthe daily workload and you know
a small number here, a smallnumber there, and I think the
same will be.
That will be the same for kindof ai agents.
(31:22):
It will be.
Certain aspects of certainroles will be.
You can completely leave it tothe agent and it's like okay,
yep, these look reasonable,let's kill these two.
So, yeah, I think that's goingto be the next big thing.
And it will be a big thingbecause there'll be a kind of
consumer facing element to it,but probably the real value will
come from the boring backendadmin stuff that will be
(31:44):
automated as a result.
Speaker 1 (31:46):
And do you see those AI agents being heavily domain
specific?
So, for example, using theexample you gave, there would be
a travel one.
Yeah, there would be aparalegal one.
You know, is that how you seeit going, or is that a
limitation I'm putting on itthat actually doesn't need to be
there?
Speaker 2 (32:05):
so I mean we've been building a as part of cortical
chart framework for doing this.
Um again it's, you know we'reat the limits of the technology
as for the back end stuff likellms, but we've been doing stuff
across travel.
I mean, ultimately you need aframework where you can break
tasks down, an llm like likechat, gpt or GPT-4.
Rather, if you give it too muchfor any one task, it's never
(32:29):
going to do a good job, like themore instructions you give it,
the less task cohesion it has.
So it's it's like you've got tobe able to break the task down
and have agents pass bits alongto other agents, almost like a
network of various more specificthings.
So if you make an element of atask specific enough, then GPT-4
(32:49):
will do a pretty good job,pretty consistently.
So it's like how do you breakdown booking a flight or sorting
out a holiday into varioustasks?
Well, you know, the tasksthemselves might be different,
but the framework to arrangethose things can be a sort of
layer that works across a bunchof different domains.
Speaker 1 (33:07):
Right and sort of changing tack slightly.
I'm interested when we weretalking before about some of the
impact that venture capital hason founders and the inability
to control your strategy, reallybecause your capital strategy
has been outsourced to somebodyelse, essentially, what's the
scene like in the UK?
A lot of the listeners of thispodcast are in Australia and New
(33:27):
Zealand.
What's the scene like aroundfounders deciding to sort of
look at some more of a publicmarket approach in the early
stages?
So they've gone past the angels, they've got friends and family
, they've done angels, but theydon't want to go down the
venture path and they want tomake sure they get to control
their capital strategy alongsidetheir business strategy.
(33:48):
Is there much of an opportunityto get into public markets
early in the UK?
So do you mean?
Speaker 2 (33:52):
like listing on a stock exchange Listings.
Yeah, I'm not sure it's thateasy in the UK.
I mean, places like Canada havea stock market for very small
cap-sized companies, so we had alook about actually ipa in
there many years ago and, butnot, I'm not so much in the uk.
I don't think I'd also saythat's maybe not the best idea
if you're a smaller companybecause there's so much more
(34:13):
oversight needed.
Yeah, the regular, you've gotall the extra.
Yeah, you got your quarterlythings on display for everyone.
Speaker 1 (34:21):
Yeah, um, yeah and what about the UK startup scene
generally?
Like is the UK startup scene,you know, compared to where it
was two or three years ago?
Is it thriving, Is it sort ofbouncing back or is it still
struggling along a little bit?
Speaker 2 (34:34):
I think that it's been quite difficult.
Due to the interest rate andeverything like that.
I think VCs and other kind offinanciers have sort of pulled
back a lot of investment simplybecause, you know, it's the
economic conditions really.
So it's been a lot harder toraise money and companies that
were kind of relying on gettingsuccessive rounds have been kind
(34:57):
of forced to do down rounds andthings like that.
So that creates a bit of avicious circle where VCs are
kind of like less scared toinvest in the first place
because they don't want to havea down round.
So yeah, it's not been ideal.
I mean, as I say, we haven'treally been courting that for a
while, so I wouldn't say we havea figure massively on the pulse
, but that's kind of what I'veheard from others who have been
(35:18):
much more involved.
It's been pretty tough.
Speaker 1 (35:20):
Yeah, it's been pretty tough.
Yeah, and you you also talkedat the start of the show about
almost a philosophy shift thatwent along the lines of you know
, you worked out early onwhether it was pre university or
sort of, at the time you're atuniversity that by working hard
and by being focused, just theoutcomes just changed completely
.
What was it that shifted foryou?
(35:40):
Because that's a big change.
Most people don't make a shiftlike that.
They're either like that orthey're not.
What was it that changed?
Was it your father and passingaway, or what was involved there
?
Speaker 2 (35:51):
I think I was like what am I doing with my life?
Do you know what I mean?
I had all this opportunity andI'm kind of squandering it.
And I felt so muchself-loathing, if you like,
about that.
I was like I really don't wantto be this person anymore and I
would sort of write a plan ateach beginning of each week.
So I'm going to do this many.
I use Pomodoro.
So this many Pomodoro's of workon my business, this many on uni
(36:13):
, this many on other things likefitness or like boxing or
whatever, and the sense that ifI didn't achieve these goals I
was sort of sliding back intothat person I didn't want to be,
was enough of a kind of mentalcounterweight.
I think maybe if I hadn't hadsuch a profound negative
experience, that wouldn't havebeen there.
So it's like having getting tothat point, I really hated
(36:36):
myself for who I was in thesense of like I felt like I
wasn't achieving what I could beachieving, not in, not in a
more kind of general sense, butreally I was letting myself down
massively.
And as soon as I started doingthat, I started to notice a big
difference.
But funnily enough, you'dafford it, and so as soon as you
actually start working hardthings, things tend to improve
(36:58):
quite dramatically.
So I was like, getting fitter,I was doing better at uni, I was
getting my more consultingstuff off the ground, and then I
was like, well, if I ever stopdoing this, I'm gonna go back to
being that person.
So it was kind of like afeedback loop of the more I
progressed, the further I felt Igot from that person almost,
the more I could lose, thefurther I had to fall, the more
(37:18):
I was more committed to notdoing that, and so I could push
myself further because I waseven more committed to this new
path, if you see what I mean.
So that was the kind of mindsetI had.
Speaker 1 (37:29):
And do you ever slip back?
Is there ever a tendency toslip back to those old you know
almost childhood habits andthoughts, or is it actually so
hardwired in you now this change?
Speaker 2 (37:39):
I think, yeah, I've experimented with this and so I
find if I, you know, if I don'twrite down like a weekly set of
things, after a few weeksobviously, just, it's a lot
easier to stay motivated whenyou've got a team and a customer
and it's not, it's not really aquestion of self-motivation
anymore, because the externalworld motivates you, but other
(37:59):
stuff outside of work, gym andso forth, that that tends to
slack off.
And I think you do, habits dochange and I think you know I'm
a million miles different anyway, but improvement does stop, I
think.
So I found I'm just going tokeep doing it.
Yeah, and it's kind of like Ihave sort of a grand plan of
what I want to do in my life andeach, each week, if you like,
(38:22):
well, I'll do a little bittowards that various aspect of
it, this aspect, a sense ofbeing on plan and going towards
a purpose, and even if you'remaking small progress, that's
quite a powerful thing formental health and sense of
wellbeing.
So if you're, you know, at theend of each week you're like,
yes, I've got a bit closer to mygoals, I think that's pretty
powerful.
And like writing down a planand achieving it is on a
(38:44):
short-term basis is a great wayto give your brain that
nourishment of like yeah, I'm onthe right path, I am getting
closer to my goals.
Speaker 1 (38:51):
I mean, we naturally want to feel we've got a purpose
and I think a lot of us want tobuild something.
So by making steps to buildthat we can see we're building
something, that we can see we'rebuilding something, it's a
powerful motivator.
And what would you say for youand I hate the phrase zone of
genius, but let's use it what isyour superpower Like?
When Alex is working in thisspace, the business is going to
(39:15):
be a lot better off becauseyou're working in your
superpower.
Speaker 2 (39:17):
What is that?
Yeah, I mean for me it'dprobably be being kind of good
at the sales and the tech.
So I mean, for me it'd probablybe being kind of good at the
sales and the tech.
So I mean a lot of businessesstruggle where you've got a
sales team that don't understandthe capabilities.
So especially when you're doingsomething like where you're
right at the cutting edge, wherethe capabilities are not well
defined, so can this be solvedwith AI?
Speaker 1 (39:39):
or not.
Speaker 2 (39:39):
It's not an easy question to answer, but if
you're trying to sell a projectwhere, okay, we want you to
automate this aspect of ourbusiness, if you can be that
salesperson that alsounderstands the capabilities,
that's a pretty strong.
You can really.
The client can obviously tellthat you know what you're
talking about and you can answerany technical questions on the
fly and you cannot commit yourcompany or your team to this
(40:01):
completely unfeasible projectand, conversely, on the backend,
on the engineering side, youcan make decisions based on well
, is this going to get us tothese objectives that matter to
the business?
But you know, there's, I think,a tendency to overengine every
engineer.
I mean, I'd like to call myselfan engineer to some degree, to
(40:24):
some degree and the tendency isto just try and make everything
really awesome and robust andlike as in line with best
practice as possible, becausethat's you kind of want to do
what you've learned.
But that isn't necessarily thebest decision from a business
point of view.
You know, okay, if we, we don'tknow if this feature is going
to actually be worth building,so let's build a quick version
of it.
Okay, that might mean we haveto, like, take these shortcuts
in the short term or maybe notdo this thing.
(40:44):
We should do first, but wemight not need to do this at all
.
So sometimes developers,divorced of the business context
, can go down rabbit holes andover engineer the things that
you don't even need.
So kind of having a foot inboth camps, maybe the business
sales side and the tech side.
I think is is really powerfuland I think that's been probably
the probably the reason I'mhere, to be honest, because you
(41:06):
know you don't need to bebrilliant like overarching
genius to both either of thosethings, as long as you know
enough to be dangerous at both,at both yeah, and is that the
sort of thing that you wouldadvise a founder on?
Speaker 1 (41:17):
is you know you need to ideally be.
You don't have to be an expertat something that you it's good
if you're, if you're okay orgood at a couple of things like
what piece of advice have youbeen imparting to other?
You know younger founders orpeople starting out over the
last few years where you'velearned something and you think
you know I wish I'd known thisand so I tell lots of founders
this particular piece of advice.
(41:39):
What would that?
Speaker 2 (41:39):
be?
Yeah, exactly, I think.
I think it would probably betwo things.
Like one would be trying to askyourself, like why am I
starting this business?
What am I actually trying toget out of it?
Like yeah, actually, but beingsuper honest with yourself, is
it?
Are you trying to change theworld or do you want to make
money?
You know, do you want toprovide for your family and have
a comfortable life?
Because those take radicallydifferent paths, you know.
(42:01):
I think one would be the v money, one would be for God's sake,
don't get VC money becauseyou're gambling everything.
And the other would be to tryand if you're a hard skills
person when I say hard skills Imean a techie person let's say
try and put yourself out thereand be in more sales and
business situations and, if youcan like, do that in a way where
(42:21):
you're kind of it's just youand a client and you, you learn
those skills because that itwill help you appreciate.
You know, whatever camp you'rein, try and understand a bit
more about the other campthere's lots of.
I've got a friend who's 37 andhe's uh, you know, always been
in sales and he's he's doinglike a data science and python
course, and things.
That's just you know.
You can do these things forfree.
It doesn't take you very manyhours and you've gone from zero
(42:43):
to way more than zero and it'squite a good investment in time
and energy and just makes youway better at appreciating what
you maybe you're asking of otherpeople when you're making a
sale or, you know, creating arequirement and things like that
.
So, yeah, that would probablybe my, my advice.
Speaker 1 (43:00):
Yeah and how do you inspire yourself?
You, I think I saw somethingabout you do some boxing.
Is that right?
You do boxing at the gym, ordid.
Speaker 2 (43:10):
Well, yeah, it's a bit of a story there.
So at school I was always a bitof a nerd, so I never really
stuck up for myself and it wasalways a part of me.
I didn't really like, I sort ofhated it, in fact, as you
probably could imagine.
And I was out in um on holidaywith my mates in Croatia and one
thing led to another and I gotstarted on by some fella in a
bar and I didn't stand up formyself at all and I think as a
(43:32):
result of that we ended upgetting piled on by like 30 guys
and I got quite badly hurt andI was like if I'd have just not
backed down in that situation,I'm fairly sure that wouldn't
have happened got a sort ofsmelt weakness, if you like.
So, like what can?
What's the single most hardcoreway I could make myself tough?
It was like right, I'm justgonna join a boxing gym and it
was incredibly tough.
(43:53):
You know, I was in the firstcouple of sparring sessions.
I was like afraid to like hit,even hit people because I just
felt so wrong.
But after a year or so, doingthat I was, I was completely
cured of that part of me andthat was actually really nice, a
really nice feeling ofsomething that's always you've
hated about yourself, kind ofsolving that and just feeling
like I'm actually not thatperson.
I could never go back to beingthat person.
(44:14):
That would not stand up forthemselves.
So that was awesome.
And I did boxing for a bunch ofyears and my dad had
Alzheimer's and both of hissisters had it and I'm like
maybe I shouldn't be getting hitin the head so much.
So I was like maybe I'll stopthat.
So currently I'm just going tothe gym a lot.
I've got ambitions to get intoBJJ and some of the wrestling
(44:34):
side of mixed martial arts youcould take on Mark Zuckerberg,
yeah exactly.
Yeah, I think I've got about 20kilos on Zach.
It'll be fine.
Speaker 1 (44:46):
Yeah, I think most humans are heavier than Mark
Zuckerberg.
You could be some sort of bigAI challenge.
That would be quite funny.
And do you find that?
It's just a complete antidoteto startup life and if you've
had a bad day, you can smack thecrap out of a bag, or you know?
I mean because you're notsparring now, presumably because
you don't want to get smackedin the head.
Speaker 2 (45:06):
Yeah, I mean, it certainly is.
I mean any kind of I think anykind of goal outside of startup
life is a great thing to haveand it doesn't need to.
Obviously it doesn't eversupersede your business, but,
like my co-founder, for example,started doing tennis, he's now
got into the advanced league andhe didn't start until with you
know, he's about eight yearsolder than me and he didn't even
start till we're like fouryears into the business and he's
(45:28):
super competitive and it's likesomething where he can you know
, progress and it's just a timeto switch off.
You know, for me at the momentit's just like I'm trying to.
I always wanted to get into likegreat shape at the gym and then
I sort of for some reason,right, I was doing this really
seriously for like a year beforefounding the business and then
my thought was like, well, we'llsell the business in like two
years and I'll be like on ayacht and I'll have like a
personal trainer.
(45:49):
It's like, obviously doesn'twork out like that.
But it was like seven years ofdoing absolutely no exercise at
all.
And then I was like, right, youknow what, I'm going to start
working towards this, eventhough I'm still doing a
business and it's like found.
You know, I'm going to the gymabout six times a week and you
know, doing all the proteinburners and so forth.
So just trying to get into goodshape has been my main goal.
(46:09):
I mean the boxing side ofthings.
I'm trying to help my cousin atthe moment train for a fight
and that's quite fun.
But, yes, I think anythingwhere you can see improvements
in yourself or some other aspectof your life that can give you
a, I think, fundamentally humanpsychology, if you see growth
somewhere, that's an incrediblynourishing thing and it's an
incredibly powerful way ofprotecting your mental health.
(46:31):
So, yeah, if you can be doinganything where you want to be
doing, it ideally a goal as well.
If yours and you're makingsmall progress towards it, then
that can be like a lifeline ofeverything's getting
overwhelming in the office.
That towards it, then that canbe like a lifeline of
everything's gettingoverwhelming in the office.
Speaker 1 (46:43):
That is a really wonderful way to finish and
that's superb advice for peoplelistening and you know Mark
Zuckerberg, if you are listening, you know Alex is ready for you
.
He is ready to fight you.
He is ready to go.
Bring it on, Bring this on.
So, Alex, has been absolutelyfantastic chatting to you today.
Thank you so much for the timeyou've given.
What I'll do is I'll make surethat I put some connections to
(47:06):
you and to Cortical in the shownotes so that if people are
interested in learning moreabout Cortical or wanting to
reach out to you personally,they can do so in the show notes
and connect with you.
Again, thank you so much foryour time.
It's been a real privilege anda real honour chatting to you
today.
So, yeah, thank you.
Thanks so much for having me.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!