October 27, 2025 • 24 mins
Spill the tea - we want to hear from you!
What if you could stress‑test a policy pitch with a faithful simulation of the minister you’re meeting tomorrow? We sit down with Leon Emirali—startup founder turned Westminster aide—who’s now building Nostrada AI, a platform that creates high‑fidelity digital twins of politicians trained only on their own public words. From late‑night Brexit votes to the first days of COVID, Leon’s time inside government shapes a clear-eyed view of how information moves, where decisions bottleneck, and why faster, more grounded research can improve the conversation between industry and state.
We dig into the mechanics: how per‑politician models reduce cross‑bias, why provenance and citations matter, and where to draw bright lines between public records and private data. We also probe the big question lighting up headlines—AI “ministers” and democratic legitimacy. Lebanon and Albania are experimenting; should the UK or US follow? Leon argues for AI as a tool, not a ruler, and lays out practical guardrails that keep expertise accessible while preventing false authority and misuse.
Beyond politics, we explore open data as a strategic asset for smarter public services—think anonymized NHS patterns, transport peaks, and service planning informed by real signals. And for builders, Leon shares hard-won advice: say yes, ship sooner than feels safe, and let feedback compound. If you care about policy, lobbying, AI ethics, or the future of democratic engagement, this conversation offers concrete insights, skeptical optimism, and a working blueprint for responsible adoption.
If this sparked ideas, follow the show, share with a friend who loves tech and policy, and leave a quick review so more curious people can find us.
Links:
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Stephen (00:02):
And welcome to another episode of The Incongruent.
This is Stephen King, andjoining me today is Rad Rad
Radhika.
Radhika (00:10):
Hi everyone, thanks for tuning in.
Stephen (00:13):
And if you are enjoying this new season, please do
like, subscribe, follow, andgive us some thumbs up or some
comments wherever it'snecessary.
It really does appreciate it.
Radhika, we had an amazingconversation today.
Who did we speak to?
Radhika (00:28):
So we spoke to Leon Emirali and we dived into his
diverse career across socialmedia marketing, politics, and
now AI.
And he told us about thiscutting-edge uh political AI
tool that he's been working onin Australia.
So tune in if you want tolisten more.
Stephen (00:46):
Yeah, we're going into synthesizing uh data based on
public servants' publishedcomments.
So we are creating digitaltwins here.
Oh no, we are, but uh Leon ismaking digital twins of
different government ministers,and you'll be able to have a
conversation uh with them usinga chatbot, which is really quite
(01:07):
cool.
Uh we go into all the ethics ofit, we go into potential uses
uh and where this might go.
So it was an absolutely amazingjob, right?
Radhika (01:15):
Yes, it really was.
Stephen (01:17):
And Radhika did a great job.
So if everybody's ready, herewe go.
Radhika (01:26):
Welcome to another episode of The Incongruent.
Today we have with us LeonEmirali.
In 2009, while still atuniversity, Leon started one of
the UK's earliest social mediamarketing companies.
He then went on to co-foundCrest, the award-winning digital
agency.
After leading the business toyear-on-year growth, his
shareholding was acquired in2019.
(01:47):
After successful exits from twostartups, he was appointed as a
chief aide to a member ofparliament and UK government
cabinet minister.
Following an eventful period inWestminster, which spanned the
UK's withdrawal from the EU andonset of the COVID-19 pandemic,
Leon returned to the privatesector advising executives who
have led global businesses,including Aston Martin, EasyJet,
(02:10):
and Virgin Galactic.
Today, outside of running hisown ventures, he works as an
advisor to advertising agencyMNC Satchie and political
consultancy, PLMR.
Leon often appears as an expertcommentator in the media and is
a regular on BBC News and SkyNews.
Leon has written columns forthe Times, Telegraph, City AM,
(02:33):
and others.
His current vendor, NostradaAI, builds high fidelity AI
simulations of global politicalactors, enabling corporate
diplomatic and defenseapplications and forecasting
future policy testingstrategies, generating
communications material, andpreparing effectively for real
world engagement.
Welcome to the podcast, Leon.
Leon (02:55):
Thank you very much for having me.
Radhika (02:58):
So let's get into it.
So the first question that we'dlike to ask you presently it's
possible you have thedistinction of being our
youngest guest on this season ofThe Incongruent.
Could you therefore share someof the many highlights of your
relatively young career, fromyour startup entrepreneur crest
to your seat of power during theCOVID-19 Brexit and back again
(03:21):
to Nostrada?
Leon (03:22):
Great.
Well, it's not it's not often Iget called young these days,
but I will I will take it at anygiven opportunity.
So thank you.
Um yeah, no, I've I've beenI've been I've had a very varied
career really, as you as youpoint out, Radhika.
So um I I I started reallythinking about business at
university.
Um I did a uh exchange trip toChina, um, so from Coventry
(03:43):
University where I studied, andI I spent sort of six, six,
seven months in China.
And uh while I was there, sortof had the idea about creating a
social media marketing company.
And this was this was back inthe day when social media was
was very fresh indeed.
No one really understood it uhin the way that we do now.
And I think being being a youngperson back then, when I was
(04:04):
really young, was a massiveadvantage because businesses
sort of knew they had to dosocial media, didn't really know
how.
Here's a young kid with alaptop, let's let's take a punt
on him.
And it sort of grew, it grew abit from there.
Um, so that was the firstventure, and and that was that
was brilliant as a universitystudent, and that was sort of
the beginnings of of Crest,which was an agency that I
(04:25):
founded um or like I co-foundedum with uh with a former
colleague, and um we we wereagain just sort of two two young
people trying our best to getclients in the digital marketing
space, um, did a really goodjob of that, and um my
shareholding was acquired aftera few years of of really hard
work of building the business,employing people, creating uh
(04:46):
revenue, um, and uh was sort ofthinking of what to do next.
And out of the blue, got a callfrom a a friend of mine who who
knew a cabinet minister, um,and uh and they said, Do you
want to come and work in uh inparliament in Westminster?
And I thought, um, yes, I do,um, because those opportunities
don't come around that often.
And uh yeah, it was anincredible opportunity just to
(05:09):
sort of be in and around thoseincredible buildings, um, you
know, sort of sitting in theback of a jaguar and pulling up
to Downing Street was one ofthose sort of pinch me moments,
but it was it was a real sort ofopportunity to see how the
world really works, and it wasan incredible time.
We had Brexit, we had COVID, wehad a 2019 general election, it
was all sort of happening atonce.
Um, so even though I only didthat job for a short period, um
(05:33):
it spanned a lot, so I couldhave been there for a long, long
time, it felt like anyway.
Um, and then went back into theprivate sector, doing some
advisory consultancy.
But as you say, Radhika, theexciting thing I'm working on
now is Nostrada AI, which is umreally sort of taking a
combination of my work inpolitics uh alongside my work in
sort of the digitalcommunication space, putting
(05:54):
that all together, and we'vebuilt a platform that ultimately
um creates digital twins ofmembers of parliament um or any
politician uh really um and itenables businesses,
organizations, governments togain political insights from the
the data repository that weprocess as part of the system?
Radhika (06:15):
Okay, it sounds like you've had like a really
interesting, really dynamiccareer so far, and from what I
can understand, it's like you'rereally good at understanding of
the potential and theopportunities that come your
way, and really good at graspingthem.
Uh, definitely something ourlisteners can take away.
Um, our next question is whatwas it about your education or
family upbringing that inspiredyou to become an entrepreneur at
(06:38):
such an early age?
Leon (06:40):
Yeah, it's a good question.
I think my my dad ran abusiness.
Um he was a sign maker and hewas he was self-employed, and
um, you know, I saw as a youngkid, you know, I saw my dad um
putting in the hours, putting inthe graft, um working for
himself, and that's all I'd everknown, really.
He did have jobs prior toworking on his own, but I didn't
(07:04):
really see them because I was Iwas either a baby or or before
I was born.
So my whole time knowing mydad's working life was him being
a business founder.
He wouldn't call himself anentrepreneur, but um, but that's
what he was what he was, uh,and he was self-employed, and
um, so I saw that, and I thinkthat set the blueprint really.
Always had an interest inbusiness for that reason, and I
think that was the inspirationto want to do something on my
(07:27):
own accord.
Um, but you know, having saidthat, I've had jobs as well
where I've been employed, um,huge valuable experience.
It's always good to make makeyour mistakes on someone else's
time and payroll um so that youdon't make them on in your in
your own ventures.
Um there's a lot of learning tobe done as an employee.
Uh, I'm not knocking that, butum, but for me, it was always
(07:48):
wanting to start my own thing,do my own thing, you know,
create my own path.
And I think that comes fromseeing my dad do that, really.
So I'd say that's that's whereit came from.
Radhika (07:57):
That's uh brilliant, Leon.
Um, and so the next question isyou attained quite a senior
position counselling a prominentminister in the post-referendum
British Parliament.
What was that like?
Leon (08:08):
Yeah, it was incredible.
Um I've as well as businessbeing an interest, I've always
had an interest in politics.
And uh, you know, theopportunity to work for a
government minister is quiterare.
Um it was it was in the it wasin the Brexit department
initially, uh, which wasinteresting because I I voted to
remain.
Um but uh you either do thisjob or you don't do this job, so
(08:32):
I sort of had to had to had tomake the best of it.
Um but it was an amazingexperience and incredibly
humbling um to be there as it asa fairly young young person,
sort of in my early 30s, Ithink, at the time.
Um, but that's the wayWestminster operates, actually.
I I certainly wasn't theyoungest um aide and advisor.
There were there were peopleyounger than me.
Um so I think that's just theway that it the way that it
(08:55):
works.
And you need to be, you need tobe energetic.
Um, there's not much time forsort of family life or or
anything like that.
Um and I think it does suityoung people as as their kind of
means of being able to move on,and maybe a lot of them will
want to become MPs themselves, alot of them will want to stand
for office themselves.
Um, but uh for me it was justabout getting that experience in
(09:16):
in Westminster.
Um, and as I say, it was it wasa pretty historic time, so we
saw a lot.
Um, and uh I think in Britishpolitics at that time, and
people probably probably don'tremember it, but it was it was
you know parliament was on TVevery night, um, Brexit votes,
and then and then as soon asthat was done, it looked like
that was done.
The COVID happened literally onthe day we left the European
(09:38):
Union, we got the first case ofCOVID.
Um, so it was bouncing from oneone sort of crisis to the next,
really.
Um so it was a it was a reallearning curve, and um so two
departments, my minister workedin department for exiting the
European Union, and then he wasreshuffled to the Treasury.
Um so again, that was a goodexperience to to see another
another department up close umand just working in Westminster,
(10:02):
being in in the chamber, beingin those buildings, it does sort
of bestow on you a sense of umsense of history, really.
So it was it was an amazingexperience that I'm uh very
grateful for.
Stephen (10:12):
Brilliant.
So they're moving on now toNostrada, which brings
everything all together.
Uh this fantastic digitaltwinning, synthesized data,
machine learned.
Tell us a little bit about whatNostrada is and why it makes it
so exciting.
Leon (10:29):
Sure.
So the the idea actually cameum back in 2018 when I was when
I was working at Crest, um, thedigital marketing agency that I
co-founded, um, I was doing alittle bit of advisory to um to
number 10 to to Downing Streetwhen Theresa May was Prime
Minister.
And we were looking at how dowe win the next battleground?
(10:49):
Because I think there was asense that the Conservative
Party at that time you knowhadn't mastered social media and
they'd sort of been been losingthat battle to the Labour
Party.
Um, so how do we then sort ofget an advantage for the next
battleground?
And I looked at what was comingnext in 2018 and thought, well,
it is AI, it is it's actuallyvoice.
I thought voice was going to bea big, a big moment, and it has
(11:10):
been with sort of things likeAlexa and and Siri.
Um, so I advised number 10 atthat point they should create a
Theresa May uh uh persona thatyou can talk to your Alexa
device and say, Hey PrimeMinister, what are you what are
your priorities for today?
What are you doing for today?
And she would tell you whatshe's up to that day.
And uh and I sort of presentedthat idea and I got laughed out
(11:33):
the room, if I'm honest withyou, at that point it was almost
like that that that that wouldnever happen.
Um uh both for technologicalreasons and political reasons,
it was it was sort of ridiculeda little bit, and there are a
couple of media articles aboutit ridiculed ridiculing it.
Um but that to me set the seedthat you can create these
personas of politicians, and thereasons why you can do that is
because they create so muchdata.
(11:55):
So whether that is speechesthey're giving in parliament,
whether it is social mediatweets, whether it is you know
media interviews, whatever itmight be, they are constantly
generating this data.
And then when sort of the theLLM uh revolution took off in
recent years, it just seemed tome that that was a natural
marriage between what we've gotwith with the data that
(12:17):
politicians produce, whichcrucially are in is in open
source, is it it's public databecause it's politicians
producing it.
Um the world owns that, itisn't anyone's anyone's uh
property.
So I thought, well, that makessense that we put that into an
LM model, into a into a machinelearning model, and make that
work as a as an insights umengine for for for politics, and
and that's where we're at.
Stephen (12:40):
That's great.
I mean, we'll we'll drill downa little bit into these um into
the creative commons that uhmany of the ministerial
documents are under, and I'msure that's going to change as
soon as they see what we couldpotentially do with this kind of
thing.
But we've seen uh Albania wasan exciting thing.
So we've had Diela.
Um Lebanon had an uheffectively one of the
(13:03):
newspapers in Lebanon uh createda AI minister to help advise
its own government.
Um what's your opinion on thosedevelopments?
And is is that somehow similarto what you've developed, or is
is is that how do you thinkabout those things?
Leon (13:18):
Yeah, I I think they're really interesting developments.
I think they are reallyinteresting.
I think we're a long way off ofdoing that in uh in the UK or
the US at any point soon.
But it's great to see othernations exploring it.
Um, you know, the difficulty isno one voted for that AI.
Um, and there's therefore aquestion about the democratic
(13:38):
deficit, is if you if you let AImake decisions for the country,
uh, I think there is there is aquestion to be asked.
Um, but I can envisage a timein the future where actually AI
is on the ballot, where theoption is do you want this guy,
do you want this woman, or doyou want this robot to be making
these decisions for you?
And uh, you know, if the if theAI model can prove that it
(14:01):
makes good decisions, then whywouldn't the public want to want
to put their faith in them?
Um but we're not there yet.
So I think we're a little wayoff.
Uh Albania and Lebanon andthose countries are definitely
um pioneers on that front.
Um, but I think there is aquestion to be asked about
whether that is democraticallythe right thing to do or not.
Uh I've got I've got some uhsome reservations about it.
Stephen (14:23):
Uh in my hometown of Hartlepool, uh they voted uh the
football mascot mayor twice.
Uh Hangus the Monkey as aprotest uh vote.
So I I can definitely see if AIcomes up onto the uh ballot
point, there will be a lot ofpeople who will say, uh, let's
(14:45):
just stuff everybody and we'resupport for this, um, which is
which is really quiteinteresting.
Um but they just the some ofthe complaints about the
Albanian government was allabout who trains this AI and
where does the data come from?
And we had a discussion a weekor so ago uh with a lady named
Monica Marquez who was talkingabout the the inherent biases
(15:05):
and we cannot avoid biases.
So even in your own materials,you how do you um avoid there
being a little Leon in all ofthese different ministers, if
that makes sense?
Leon (15:20):
Yeah, there's a really important question that that
sort of was was a realconsideration creating this
model, because um, you know, Ithink everyone has a bias to
some degree, uh, whether theylike to admit it or not,
naturally.
Um, but what we've done withNostrada is we have trained each
of the individual models, sothe 650 MPs, we've trained each
of those models only on the datathat that MP has generated.
(15:43):
So if you spoke to the NigelFarage digital twin, you would
get some quite right-wing views.
If you spoke to the JeremyCorbyn uh digital twin, you
would get some very left-wingviews.
And I think that's the beautyof this model is that it
reflects only what thepolitician um has inputted and
it synthesizes that data.
So obviously there are there issome system prompting that goes
(16:06):
over the top of that.
Um, but but we are very stricton saying that you should you
should accurately, as best asyou can, reflect what has been
said previously by thesepoliticians to make your to make
your case.
So it's not like a j anoff-the-shelf LLM where you
might chat to you chat to chatGPT or Gemini or whatever, and
you might sense a bit ofpolitical bias because they're
(16:28):
they're coming at things throughone lens.
This comes at things through650 different lenses, um, each
of what that MP's own ideologyis.
So we so we we try to reflectthat, and I think we we've done
a pretty good job of that um bysort of testing those those two
extremes in parliament, who areyour right-wing MPs, who are
your left-wing MPs, um, and youdo get very different answers
from each of them, as you wouldexpect, because that's the data
(16:50):
that it's been trained on.
Radhika (16:51):
So um on the Nostrada website, you identify the three
main users as chatting with AIsimulations of politicians,
predicting their future actions,and generating bespoke
communication material.
Uh, this is an ethical mindfeel on so many levels.
So, what was it that inspiredyou to tackle such a challenge?
Leon (17:11):
Yeah, I mean to be honest, I I think that AI ethics is
something that is obviously ahot topic right now, um, and and
rightly so, because it's a newtechnology that we are getting
to grips with in many ways.
Um and with this, you know, weare we are aiming this at quite
sophisticated enterprise users.
(17:32):
So these are people who work ingovernment relations, they
might be lobbyists, uh, theymight be a government, they
might be civil servants who whohave used the platform.
Um, you know, that theyunderstand politics and they
understand government quitewell.
Um, so I felt more comfortableabout providing them with these
tools as a means of being ableto do their job better and
(17:52):
quicker.
Um, whereas I would feel lesscomfortable if this fell into
the hands of you know youraverage sort of Tom Dick and
Harry voter who who maybe aren'tas politically sophisticated,
who would perhaps ask thechatbot a question and take it
as read that that is the exactview of the minister or whatever
it might be, despite all thedisclaimers that we have put on
(18:13):
the website, and you know, it'squite clear that it's it's not
their views, it's a digitalsimulation of their views.
Um, so I I I wouldn't becomfortable with necessarily
putting it into the hands of thepublic, but in this sense, it
is enterprise users whounderstand politics, who get
politics, um, and they're doingthis stuff anyway.
I mean, it's basically a alayer of research, advanced
research, so that that thatwould take them many hours to do
(18:35):
manually.
We're we're making we'reallowing them to do it a lot
quicker and a lot better umthrough the way in which that
we've packaged this.
So um, so that that was mythought process behind the
ethics of it.
Um obviously the ethics aroundthe the data, so that was a
consideration, and and the greatnews that we we've already
touched on is that the greatnews for us is that the data is
in common common uh use, soanyone can sort of access that
(18:59):
data and and and use it.
It's not the MP's intellectualproperty.
That's what you get for being apublic servant, I guess.
Um we don't use any of theirprivate data, so even if we
know, I don't know, let's sayI'm friends with one of the MPs
on Facebook behind behind a umyou know a private account, we
wouldn't input that data ontothe platform.
It is only data that is alreadyin the public domain that we
are that we are synthesizing.
(19:20):
Um so there were a lot ofquestions about ethics, big
consideration for us, but wewe've got to a point where we're
quite happy with with with whatwe're able to produce, and we
think is actually you know do itdoing quite good for the world
and being able to enablebusinesses to better understand
how policy is made, how they caninfluence policy in a positive
way.
Um and uh and and and thatseems to be working quite well
(19:42):
for us.
Radhika (19:46):
Yeah, so you touched on this a little bit already, and
there are some conflicting umpoints on whether putting this
information into the hands ofcorporate elite groups and
lobbyists and possibly consumersis a democratic service or
apparel.
So what is your perspective onthat?
Leon (20:03):
Yeah, look, look, lobbyists get a bad rap, um, and
I get that because that thatthat's where I started my
career.
Um, but I think if you if youlook beyond uh the headlines,
and you know, it isn't sort ofbrown paper envelopes and and
grubby deals, it's actuallyeducating.
It's about it's about educatingMPs and ministers who by and
(20:23):
large tend to be generalists,and very few of them are
specialists in a certain topicarea.
So if you're the minister incharge of health policy, let's
say, um, wouldn't it make sensethat you are speaking to a
company in health who areexperts on, you know, whatever
it might be, a certainpharmaceutical or whatever?
Um, wouldn't it make sense youspeak to them, you get to
understand what they're tryingto do, so you can make policy um
(20:46):
in the most informed way aspossible.
And that's what this is,effectively.
It's giving companies a betterway of engaging with
parliamentarians who who aren'tspecialists.
Um so we're trying our verybest to enable that process to
make people feel comfortable inthe policies that they are
delivering.
Um, and uh and that's whatwe're doing.
You know, yes, there is a youhave to pay for this, you have
(21:08):
to you have to um you have tomake money to use this.
Um I understand that, but itisn't necessarily a critical
function.
We aren't saying that you can'tengage with your MP in the
normal way as as a as a voter,as a normal punter, we're just
providing this as an enterprisetool to help businesses um
engage more effectively, and Ithink that's better for
everyone.
Stephen (21:30):
Moving towards the end here, I mean we've we've had a
few of our guests talking aboutthis public uh data, data that
is already in the public domainand which is very rich.
Um given that you are alreadyin this field, uh have you
knocked heads with other peoplewho are uh monetizing other data
(21:51):
points?
Um what or or do you think thatthere are other pots of data
gold out there that we that wecould be looking at if we wanted
to accept our own?
Leon (22:01):
Yeah, it's a it's a that is a massive question because I
think that's going to underpin alot of what happens in AI in
the next few years.
And certainly in the UK,there's legisl legislation going
through Parliament right nowaround what do we do with with
AI.
Is it a de uh what is it sowhat do we do with content and
AI training?
I think the question at themoment is do we make it a
default um that that anythingthat's put out there AI can be
(22:25):
trained on.
Um, you know, and I think thatthat's come under a lot of
scrutiny from musicians andcontent creators, obviously,
because they want to they wantto be paid for what they create,
and I think that's a fairpoint.
Um, but no, I think there is alot of of of rich open data, as
you say, Steve.
Um I I I haven't knocked headswith anyone doing anything
similar, to be honest.
I think because it's quiteearly in the in the process
(22:46):
here.
Not many people are are sort ofyou know in this space
necessarily.
Um but I do think that thereare opportunities for us to look
at what what's availablepublicly and how can we how can
we synthesize that.
I think the government inparticular should be looking at
it um in terms of what data theyhave access to.
So for example, NHS records, ahuge amount of data the
(23:07):
government owns.
If the government can use that,use AI to properly understand
that data, to to to to betterunderstand how I don't know what
time of day most people callthe GP surgery or what most what
what illnesses peak at whatpoint in the year.
If they can use that to planpublic services, I think there
is huge opportunity there.
Um, but it's about having thetool set, having the skill set
(23:30):
within government to make surethat that can actually happen.
Stephen (23:33):
Good.
I've got one last questionbecause I know you've got a hard
finish in a moment.
If you went back to youruniversity days, is there any
advice you would offer to youryounger self?
Leon (23:43):
Yeah, I I think um just get out there and meet as many
people as you can.
Say yes to as many things thatyou can.
Um if you've got an idea forsomething, just go and do it.
Don't worry about it beingperfect.
Put something out there in theworld that isn't necessarily
perfect and work on it as yougo.
Um, I think the biggestrestriction, certainly I felt it
(24:05):
as a student, was that youknow, oh, I didn't really know
how to make that happen.
Or or or it just seemed as ifthat it wasn't as simple as it
was, it was too complicated.
Just start working on stuff andput it out there, and you will
learn as you go and take everyopportunity that that comes your
way, I think is the advice I'dgive I'd give younger self.
And don't use so much hair gel,otherwise you end up with not
(24:26):
much hair on your head.
Radhika (24:27):
Yeah, so we had another interesting episode on AI and
use of AI in politics.
And uh thank you so much, Leon,for being on our podcast and
giving us all your interestinginsights, and your advice to
your younger self will also bevery useful to our younger
listeners.
Um, so join in for our nextepisode and thank you for
(24:52):
listening.
Leon (24:53):
Thanks for having me.
And welcome to another episode of The Incongruent.
This is Stephen King, andjoining me today is Rad Rad
Radhika.
Radhika (00:10):
Hi everyone, thanks for tuning in.
Stephen (00:13):
And if you are enjoying this new season, please do
like, subscribe, follow, andgive us some thumbs up or some
comments wherever it'snecessary.
It really does appreciate it.
Radhika, we had an amazingconversation today.
Who did we speak to?
Radhika (00:28):
So we spoke to Leon Emirali and we dived into his
diverse career across socialmedia marketing, politics, and
now AI.
And he told us about thiscutting-edge uh political AI
tool that he's been working onin Australia.
So tune in if you want tolisten more.
Stephen (00:46):
Yeah, we're going into synthesizing uh data based on
public servants' publishedcomments.
So we are creating digitaltwins here.
Oh no, we are, but uh Leon ismaking digital twins of
different government ministers,and you'll be able to have a
conversation uh with them usinga chatbot, which is really quite
(01:07):
cool.
Uh we go into all the ethics ofit, we go into potential uses
uh and where this might go.
So it was an absolutely amazingjob, right?
Radhika (01:15):
Yes, it really was.
Stephen (01:17):
And Radhika did a great job.
So if everybody's ready, herewe go.
Radhika (01:26):
Welcome to another episode of The Incongruent.
Today we have with us LeonEmirali.
In 2009, while still atuniversity, Leon started one of
the UK's earliest social mediamarketing companies.
He then went on to co-foundCrest, the award-winning digital
agency.
After leading the business toyear-on-year growth, his
shareholding was acquired in2019.
(01:47):
After successful exits from twostartups, he was appointed as a
chief aide to a member ofparliament and UK government
cabinet minister.
Following an eventful period inWestminster, which spanned the
UK's withdrawal from the EU andonset of the COVID-19 pandemic,
Leon returned to the privatesector advising executives who
have led global businesses,including Aston Martin, EasyJet,
(02:10):
and Virgin Galactic.
Today, outside of running hisown ventures, he works as an
advisor to advertising agencyMNC Satchie and political
consultancy, PLMR.
Leon often appears as an expertcommentator in the media and is
a regular on BBC News and SkyNews.
Leon has written columns forthe Times, Telegraph, City AM,
(02:33):
and others.
His current vendor, NostradaAI, builds high fidelity AI
simulations of global politicalactors, enabling corporate
diplomatic and defenseapplications and forecasting
future policy testingstrategies, generating
communications material, andpreparing effectively for real
world engagement.
Welcome to the podcast, Leon.
Leon (02:55):
Thank you very much for having me.
Radhika (02:58):
So let's get into it.
So the first question that we'dlike to ask you presently it's
possible you have thedistinction of being our
youngest guest on this season ofThe Incongruent.
Could you therefore share someof the many highlights of your
relatively young career, fromyour startup entrepreneur crest
to your seat of power during theCOVID-19 Brexit and back again
(03:21):
to Nostrada?
Leon (03:22):
Great.
Well, it's not it's not often Iget called young these days,
but I will I will take it at anygiven opportunity.
So thank you.
Um yeah, no, I've I've beenI've been I've had a very varied
career really, as you as youpoint out, Radhika.
So um I I I started reallythinking about business at
university.
Um I did a uh exchange trip toChina, um, so from Coventry
(03:43):
University where I studied, andI I spent sort of six, six,
seven months in China.
And uh while I was there, sortof had the idea about creating a
social media marketing company.
And this was this was back inthe day when social media was
was very fresh indeed.
No one really understood it uhin the way that we do now.
And I think being being a youngperson back then, when I was
(04:04):
really young, was a massiveadvantage because businesses
sort of knew they had to dosocial media, didn't really know
how.
Here's a young kid with alaptop, let's let's take a punt
on him.
And it sort of grew, it grew abit from there.
Um, so that was the firstventure, and and that was that
was brilliant as a universitystudent, and that was sort of
the beginnings of of Crest,which was an agency that I
(04:25):
founded um or like I co-foundedum with uh with a former
colleague, and um we we wereagain just sort of two two young
people trying our best to getclients in the digital marketing
space, um, did a really goodjob of that, and um my
shareholding was acquired aftera few years of of really hard
work of building the business,employing people, creating uh
(04:46):
revenue, um, and uh was sort ofthinking of what to do next.
And out of the blue, got a callfrom a a friend of mine who who
knew a cabinet minister, um,and uh and they said, Do you
want to come and work in uh inparliament in Westminster?
And I thought, um, yes, I do,um, because those opportunities
don't come around that often.
And uh yeah, it was anincredible opportunity just to
(05:09):
sort of be in and around thoseincredible buildings, um, you
know, sort of sitting in theback of a jaguar and pulling up
to Downing Street was one ofthose sort of pinch me moments,
but it was it was a real sort ofopportunity to see how the
world really works, and it wasan incredible time.
We had Brexit, we had COVID, wehad a 2019 general election, it
was all sort of happening atonce.
Um, so even though I only didthat job for a short period, um
(05:33):
it spanned a lot, so I couldhave been there for a long, long
time, it felt like anyway.
Um, and then went back into theprivate sector, doing some
advisory consultancy.
But as you say, Radhika, theexciting thing I'm working on
now is Nostrada AI, which is umreally sort of taking a
combination of my work inpolitics uh alongside my work in
sort of the digitalcommunication space, putting
(05:54):
that all together, and we'vebuilt a platform that ultimately
um creates digital twins ofmembers of parliament um or any
politician uh really um and itenables businesses,
organizations, governments togain political insights from the
the data repository that weprocess as part of the system?
Radhika (06:15):
Okay, it sounds like you've had like a really
interesting, really dynamiccareer so far, and from what I
can understand, it's like you'rereally good at understanding of
the potential and theopportunities that come your
way, and really good at graspingthem.
Uh, definitely something ourlisteners can take away.
Um, our next question is whatwas it about your education or
family upbringing that inspiredyou to become an entrepreneur at
(06:38):
such an early age?
Leon (06:40):
Yeah, it's a good question.
I think my my dad ran abusiness.
Um he was a sign maker and hewas he was self-employed, and
um, you know, I saw as a youngkid, you know, I saw my dad um
putting in the hours, putting inthe graft, um working for
himself, and that's all I'd everknown, really.
He did have jobs prior toworking on his own, but I didn't
(07:04):
really see them because I was Iwas either a baby or or before
I was born.
So my whole time knowing mydad's working life was him being
a business founder.
He wouldn't call himself anentrepreneur, but um, but that's
what he was what he was, uh,and he was self-employed, and
um, so I saw that, and I thinkthat set the blueprint really.
Always had an interest inbusiness for that reason, and I
think that was the inspirationto want to do something on my
(07:27):
own accord.
Um, but you know, having saidthat, I've had jobs as well
where I've been employed, um,huge valuable experience.
It's always good to make makeyour mistakes on someone else's
time and payroll um so that youdon't make them on in your in
your own ventures.
Um there's a lot of learning tobe done as an employee.
Uh, I'm not knocking that, butum, but for me, it was always
(07:48):
wanting to start my own thing,do my own thing, you know,
create my own path.
And I think that comes fromseeing my dad do that, really.
So I'd say that's that's whereit came from.
Radhika (07:57):
That's uh brilliant, Leon.
Um, and so the next question isyou attained quite a senior
position counselling a prominentminister in the post-referendum
British Parliament.
What was that like?
Leon (08:08):
Yeah, it was incredible.
Um I've as well as businessbeing an interest, I've always
had an interest in politics.
And uh, you know, theopportunity to work for a
government minister is quiterare.
Um it was it was in the it wasin the Brexit department
initially, uh, which wasinteresting because I I voted to
remain.
Um but uh you either do thisjob or you don't do this job, so
(08:32):
I sort of had to had to had tomake the best of it.
Um but it was an amazingexperience and incredibly
humbling um to be there as it asa fairly young young person,
sort of in my early 30s, Ithink, at the time.
Um, but that's the wayWestminster operates, actually.
I I certainly wasn't theyoungest um aide and advisor.
There were there were peopleyounger than me.
Um so I think that's just theway that it the way that it
(08:55):
works.
And you need to be, you need tobe energetic.
Um, there's not much time forsort of family life or or
anything like that.
Um and I think it does suityoung people as as their kind of
means of being able to move on,and maybe a lot of them will
want to become MPs themselves, alot of them will want to stand
for office themselves.
Um, but uh for me it was justabout getting that experience in
(09:16):
in Westminster.
Um, and as I say, it was it wasa pretty historic time, so we
saw a lot.
Um, and uh I think in Britishpolitics at that time, and
people probably probably don'tremember it, but it was it was
you know parliament was on TVevery night, um, Brexit votes,
and then and then as soon asthat was done, it looked like
that was done.
The COVID happened literally onthe day we left the European
(09:38):
Union, we got the first case ofCOVID.
Um, so it was bouncing from oneone sort of crisis to the next,
really.
Um so it was a it was a reallearning curve, and um so two
departments, my minister workedin department for exiting the
European Union, and then he wasreshuffled to the Treasury.
Um so again, that was a goodexperience to to see another
another department up close umand just working in Westminster,
(10:02):
being in in the chamber, beingin those buildings, it does sort
of bestow on you a sense of umsense of history, really.
So it was it was an amazingexperience that I'm uh very
grateful for.
Stephen (10:12):
Brilliant.
So they're moving on now toNostrada, which brings
everything all together.
Uh this fantastic digitaltwinning, synthesized data,
machine learned.
Tell us a little bit about whatNostrada is and why it makes it
so exciting.
Leon (10:29):
Sure.
So the the idea actually cameum back in 2018 when I was when
I was working at Crest, um, thedigital marketing agency that I
co-founded, um, I was doing alittle bit of advisory to um to
number 10 to to Downing Streetwhen Theresa May was Prime
Minister.
And we were looking at how dowe win the next battleground?
(10:49):
Because I think there was asense that the Conservative
Party at that time you knowhadn't mastered social media and
they'd sort of been been losingthat battle to the Labour
Party.
Um, so how do we then sort ofget an advantage for the next
battleground?
And I looked at what was comingnext in 2018 and thought, well,
it is AI, it is it's actuallyvoice.
I thought voice was going to bea big, a big moment, and it has
(11:10):
been with sort of things likeAlexa and and Siri.
Um, so I advised number 10 atthat point they should create a
Theresa May uh uh persona thatyou can talk to your Alexa
device and say, Hey PrimeMinister, what are you what are
your priorities for today?
What are you doing for today?
And she would tell you whatshe's up to that day.
And uh and I sort of presentedthat idea and I got laughed out
(11:33):
the room, if I'm honest withyou, at that point it was almost
like that that that that wouldnever happen.
Um uh both for technologicalreasons and political reasons,
it was it was sort of ridiculeda little bit, and there are a
couple of media articles aboutit ridiculed ridiculing it.
Um but that to me set the seedthat you can create these
personas of politicians, and thereasons why you can do that is
because they create so muchdata.
(11:55):
So whether that is speechesthey're giving in parliament,
whether it is social mediatweets, whether it is you know
media interviews, whatever itmight be, they are constantly
generating this data.
And then when sort of the theLLM uh revolution took off in
recent years, it just seemed tome that that was a natural
marriage between what we've gotwith with the data that
(12:17):
politicians produce, whichcrucially are in is in open
source, is it it's public databecause it's politicians
producing it.
Um the world owns that, itisn't anyone's anyone's uh
property.
So I thought, well, that makessense that we put that into an
LM model, into a into a machinelearning model, and make that
work as a as an insights umengine for for for politics, and
and that's where we're at.
Stephen (12:40):
That's great.
I mean, we'll we'll drill downa little bit into these um into
the creative commons that uhmany of the ministerial
documents are under, and I'msure that's going to change as
soon as they see what we couldpotentially do with this kind of
thing.
But we've seen uh Albania wasan exciting thing.
So we've had Diela.
Um Lebanon had an uheffectively one of the
(13:03):
newspapers in Lebanon uh createda AI minister to help advise
its own government.
Um what's your opinion on thosedevelopments?
And is is that somehow similarto what you've developed, or is
is is that how do you thinkabout those things?
Leon (13:18):
Yeah, I I think they're really interesting developments.
I think they are reallyinteresting.
I think we're a long way off ofdoing that in uh in the UK or
the US at any point soon.
But it's great to see othernations exploring it.
Um, you know, the difficulty isno one voted for that AI.
Um, and there's therefore aquestion about the democratic
(13:38):
deficit, is if you if you let AImake decisions for the country,
uh, I think there is there is aquestion to be asked.
Um, but I can envisage a timein the future where actually AI
is on the ballot, where theoption is do you want this guy,
do you want this woman, or doyou want this robot to be making
these decisions for you?
And uh, you know, if the if theAI model can prove that it
(14:01):
makes good decisions, then whywouldn't the public want to want
to put their faith in them?
Um but we're not there yet.
So I think we're a little wayoff.
Uh Albania and Lebanon andthose countries are definitely
um pioneers on that front.
Um, but I think there is aquestion to be asked about
whether that is democraticallythe right thing to do or not.
Uh I've got I've got some uhsome reservations about it.
Stephen (14:23):
Uh in my hometown of Hartlepool, uh they voted uh the
football mascot mayor twice.
Uh Hangus the Monkey as aprotest uh vote.
So I I can definitely see if AIcomes up onto the uh ballot
point, there will be a lot ofpeople who will say, uh, let's
(14:45):
just stuff everybody and we'resupport for this, um, which is
which is really quiteinteresting.
Um but they just the some ofthe complaints about the
Albanian government was allabout who trains this AI and
where does the data come from?
And we had a discussion a weekor so ago uh with a lady named
Monica Marquez who was talkingabout the the inherent biases
(15:05):
and we cannot avoid biases.
So even in your own materials,you how do you um avoid there
being a little Leon in all ofthese different ministers, if
that makes sense?
Leon (15:20):
Yeah, there's a really important question that that
sort of was was a realconsideration creating this
model, because um, you know, Ithink everyone has a bias to
some degree, uh, whether theylike to admit it or not,
naturally.
Um, but what we've done withNostrada is we have trained each
of the individual models, sothe 650 MPs, we've trained each
of those models only on the datathat that MP has generated.
(15:43):
So if you spoke to the NigelFarage digital twin, you would
get some quite right-wing views.
If you spoke to the JeremyCorbyn uh digital twin, you
would get some very left-wingviews.
And I think that's the beautyof this model is that it
reflects only what thepolitician um has inputted and
it synthesizes that data.
So obviously there are there issome system prompting that goes
(16:06):
over the top of that.
Um, but but we are very stricton saying that you should you
should accurately, as best asyou can, reflect what has been
said previously by thesepoliticians to make your to make
your case.
So it's not like a j anoff-the-shelf LLM where you
might chat to you chat to chatGPT or Gemini or whatever, and
you might sense a bit ofpolitical bias because they're
(16:28):
they're coming at things throughone lens.
This comes at things through650 different lenses, um, each
of what that MP's own ideologyis.
So we so we we try to reflectthat, and I think we we've done
a pretty good job of that um bysort of testing those those two
extremes in parliament, who areyour right-wing MPs, who are
your left-wing MPs, um, and youdo get very different answers
from each of them, as you wouldexpect, because that's the data
(16:50):
that it's been trained on.
Radhika (16:51):
So um on the Nostrada website, you identify the three
main users as chatting with AIsimulations of politicians,
predicting their future actions,and generating bespoke
communication material.
Uh, this is an ethical mindfeel on so many levels.
So, what was it that inspiredyou to tackle such a challenge?
Leon (17:11):
Yeah, I mean to be honest, I I think that AI ethics is
something that is obviously ahot topic right now, um, and and
rightly so, because it's a newtechnology that we are getting
to grips with in many ways.
Um and with this, you know, weare we are aiming this at quite
sophisticated enterprise users.
(17:32):
So these are people who work ingovernment relations, they
might be lobbyists, uh, theymight be a government, they
might be civil servants who whohave used the platform.
Um, you know, that theyunderstand politics and they
understand government quitewell.
Um, so I felt more comfortableabout providing them with these
tools as a means of being ableto do their job better and
(17:52):
quicker.
Um, whereas I would feel lesscomfortable if this fell into
the hands of you know youraverage sort of Tom Dick and
Harry voter who who maybe aren'tas politically sophisticated,
who would perhaps ask thechatbot a question and take it
as read that that is the exactview of the minister or whatever
it might be, despite all thedisclaimers that we have put on
(18:13):
the website, and you know, it'squite clear that it's it's not
their views, it's a digitalsimulation of their views.
Um, so I I I wouldn't becomfortable with necessarily
putting it into the hands of thepublic, but in this sense, it
is enterprise users whounderstand politics, who get
politics, um, and they're doingthis stuff anyway.
I mean, it's basically a alayer of research, advanced
research, so that that thatwould take them many hours to do
(18:35):
manually.
We're we're making we'reallowing them to do it a lot
quicker and a lot better umthrough the way in which that
we've packaged this.
So um, so that that was mythought process behind the
ethics of it.
Um obviously the ethics aroundthe the data, so that was a
consideration, and and the greatnews that we we've already
touched on is that the greatnews for us is that the data is
in common common uh use, soanyone can sort of access that
(18:59):
data and and and use it.
It's not the MP's intellectualproperty.
That's what you get for being apublic servant, I guess.
Um we don't use any of theirprivate data, so even if we
know, I don't know, let's sayI'm friends with one of the MPs
on Facebook behind behind a umyou know a private account, we
wouldn't input that data ontothe platform.
It is only data that is alreadyin the public domain that we
are that we are synthesizing.
(19:20):
Um so there were a lot ofquestions about ethics, big
consideration for us, but wewe've got to a point where we're
quite happy with with with whatwe're able to produce, and we
think is actually you know do itdoing quite good for the world
and being able to enablebusinesses to better understand
how policy is made, how they caninfluence policy in a positive
way.
Um and uh and and and thatseems to be working quite well
(19:42):
for us.
Radhika (19:46):
Yeah, so you touched on this a little bit already, and
there are some conflicting umpoints on whether putting this
information into the hands ofcorporate elite groups and
lobbyists and possibly consumersis a democratic service or
apparel.
So what is your perspective onthat?
Leon (20:03):
Yeah, look, look, lobbyists get a bad rap, um, and
I get that because that thatthat's where I started my
career.
Um, but I think if you if youlook beyond uh the headlines,
and you know, it isn't sort ofbrown paper envelopes and and
grubby deals, it's actuallyeducating.
It's about it's about educatingMPs and ministers who by and
(20:23):
large tend to be generalists,and very few of them are
specialists in a certain topicarea.
So if you're the minister incharge of health policy, let's
say, um, wouldn't it make sensethat you are speaking to a
company in health who areexperts on, you know, whatever
it might be, a certainpharmaceutical or whatever?
Um, wouldn't it make sense youspeak to them, you get to
understand what they're tryingto do, so you can make policy um
(20:46):
in the most informed way aspossible.
And that's what this is,effectively.
It's giving companies a betterway of engaging with
parliamentarians who who aren'tspecialists.
Um so we're trying our verybest to enable that process to
make people feel comfortable inthe policies that they are
delivering.
Um, and uh and that's whatwe're doing.
You know, yes, there is a youhave to pay for this, you have
(21:08):
to you have to um you have tomake money to use this.
Um I understand that, but itisn't necessarily a critical
function.
We aren't saying that you can'tengage with your MP in the
normal way as as a as a voter,as a normal punter, we're just
providing this as an enterprisetool to help businesses um
engage more effectively, and Ithink that's better for
everyone.
Stephen (21:30):
Moving towards the end here, I mean we've we've had a
few of our guests talking aboutthis public uh data, data that
is already in the public domainand which is very rich.
Um given that you are alreadyin this field, uh have you
knocked heads with other peoplewho are uh monetizing other data
(21:51):
points?
Um what or or do you think thatthere are other pots of data
gold out there that we that wecould be looking at if we wanted
to accept our own?
Leon (22:01):
Yeah, it's a it's a that is a massive question because I
think that's going to underpin alot of what happens in AI in
the next few years.
And certainly in the UK,there's legisl legislation going
through Parliament right nowaround what do we do with with
AI.
Is it a de uh what is it sowhat do we do with content and
AI training?
I think the question at themoment is do we make it a
default um that that anythingthat's put out there AI can be
(22:25):
trained on.
Um, you know, and I think thatthat's come under a lot of
scrutiny from musicians andcontent creators, obviously,
because they want to they wantto be paid for what they create,
and I think that's a fairpoint.
Um, but no, I think there is alot of of of rich open data, as
you say, Steve.
Um I I I haven't knocked headswith anyone doing anything
similar, to be honest.
I think because it's quiteearly in the in the process
(22:46):
here.
Not many people are are sort ofyou know in this space
necessarily.
Um but I do think that thereare opportunities for us to look
at what what's availablepublicly and how can we how can
we synthesize that.
I think the government inparticular should be looking at
it um in terms of what data theyhave access to.
So for example, NHS records, ahuge amount of data the
(23:07):
government owns.
If the government can use that,use AI to properly understand
that data, to to to to betterunderstand how I don't know what
time of day most people callthe GP surgery or what most what
what illnesses peak at whatpoint in the year.
If they can use that to planpublic services, I think there
is huge opportunity there.
Um, but it's about having thetool set, having the skill set
(23:30):
within government to make surethat that can actually happen.
Stephen (23:33):
Good.
I've got one last questionbecause I know you've got a hard
finish in a moment.
If you went back to youruniversity days, is there any
advice you would offer to youryounger self?
Leon (23:43):
Yeah, I I think um just get out there and meet as many
people as you can.
Say yes to as many things thatyou can.
Um if you've got an idea forsomething, just go and do it.
Don't worry about it beingperfect.
Put something out there in theworld that isn't necessarily
perfect and work on it as yougo.
Um, I think the biggestrestriction, certainly I felt it
(24:05):
as a student, was that youknow, oh, I didn't really know
how to make that happen.
Or or or it just seemed as ifthat it wasn't as simple as it
was, it was too complicated.
Just start working on stuff andput it out there, and you will
learn as you go and take everyopportunity that that comes your
way, I think is the advice I'dgive I'd give younger self.
And don't use so much hair gel,otherwise you end up with not
(24:26):
much hair on your head.
Radhika (24:27):
Yeah, so we had another interesting episode on AI and
use of AI in politics.
And uh thank you so much, Leon,for being on our podcast and
giving us all your interestinginsights, and your advice to
your younger self will also bevery useful to our younger
listeners.
Um, so join in for our nextepisode and thank you for
(24:52):
listening.
Leon (24:53):
Thanks for having me.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by Audiochuck Media Company.
The Brothers Ortiz
The Brothers Ortiz is the story of two brothers–both successful, but in very different ways. Gabe Ortiz becomes a third-highest ranking officer in all of Texas while his younger brother Larry climbs the ranks in Puro Tango Blast, a notorious Texas Prison gang. Gabe doesn’t know all the details of his brother’s nefarious dealings, and he’s made a point not to ask, to protect their relationship. But when Larry is murdered during a home invasion in a rented beach house, Gabe has no choice but to look into what happened that night. To solve Larry’s murder, Gabe, and the whole Ortiz family, must ask each other tough questions.