June 28, 2022 • 28 mins
In this episode, Somil Gupta (Founder Algorithmic Scale), Kieran Gilmurray (Author & Digital Transformation Expert), and Andreas Welsch discuss how business leaders can win with AI. Somil and Kieran share their insights on what it means to digitally transform an organization and provide valuable tips for listeners looking to lead AI programs to success.
Key topics:
- Lead digital transformation programs
- Monetize AI products & insights
- Incorporate ethics from the start
Listen to the full episode to hear how you can:
- Take a holistic view across people, place, and platform
- Increase revenue from AI when approached as a product
- Make AI ethics a part of transformation and monetization projects
Watch this episode on YouTube: https://youtu.be/VculKvDros8
Questions or suggestions? Send me a Text Message.
***********
Disclaimer: Views are the participants’ own and do not represent those of any participant’s past, present, or future employers. Participation in this event is independent of any potential business relationship (past, present, or future) between the participants or between their employers.
Level up your AI Leadership game with the AI Leadership Handbook:
https://www.aileadershiphandbook.com
More details:
https://www.intelligence-briefing.com
All episodes:
https://www.intelligence-briefing.com/podcast
Get a weekly thought-provoking post in your inbox:
https://www.intelligence-briefing.com/newsletter
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Andreas Welsch (00:00):
Today we have a special episode for you.
Now, I've been playing one of myfavorite retro game:"What's the
BUZZ?" But I'm stuck.
And that's why I've asked a fewexperts to help me"win with AI",
Somil Gupta and KieranGilmurray.
Thanks for joining.
Kieran Gilmurray (00:17):
Great to be here.
I've been in business technologyfor about the last 27 years.
I've described both because Ilove technology, but love what
it can do for a business.
I've sat in intelligentautomation roles.
I've ran AI teams for 13, 14years.
I've ran businesses, large,small, international, you name
it, I've been there, done that.
Hopefully today I can add somevalue.
(00:37):
Recently I've written a book onthe A to Z to digital
transformation.
I'm going to use some of theanswers in that book, hopefully
in today's session as well.
Delighted to be here.
Somil Gupta (00:46):
Thank you.
My name is Somil Gupta.
I'm the founder of AlgorithmicScale.
It's a company based in Sweden,and we focus on building the
monetization and commercialstack for AI based business
models.
What it means is we help ourclients build production
services.
We help them build digitaloperating models for delivering
(01:07):
the value that they create fromfrom AI.
But also help them with valuecare, which is helping them with
pricing strategies, helping themwith different modernization
strategies.
And this is something which kindof very exciting.
So I am very crazy aboutalgorithmic operating models and
algorithmic business models.
And now this company issomething that will intended to
(01:30):
make those visions into reality.
Andreas Welsch (01:34):
Awesome.
Thank you so much.
If you're just joining thisstream, drop a comment in the
chat, what tricks you arelooking for.
No cheating allowed by the way.
So what do you say?
Should we start playing?
Kieran Gilmurray (01:48):
Oh, yes,
Andreas Welsch (01:50):
Okay, so this one is a warmup.
When I hit the buzzer, you'llsee a sentence.
I need your help.
So can you answer with the firstthing that comes to mind in why,
in your own words?
So you only have 60 seconds foryour answer.
And for those of you watching uslive, please drop yours in the
(02:12):
chat as well.
Team, are you ready for What'sthe BUZZ?
If AI were city, which would itbe?
60 seconds.
Go.
Kieran Gilmurray (02:25):
Oh wow.
Let me say Boston for a coupleof reasons.
What an amazing practical city.
There is 1,001 amazing things.
There are all wrapped up in areasonably small package, and
when everybody goes to Boston,they all experience something
different.
Everybody thinks of it as adifferent place.
So I think it was a us Irishcity.
(02:45):
I'm sure a lot of other peopleremembered from all sorts of
movies and different things, buta really practical, thriving
city where a lot of tech exists,where a lot of people are making
an impact and there's still lotsof potential for things crazy
great things to happen.
Somil Gupta (03:01):
All right.
I would say Stockholm.
And reason being, I think one ofthe things which I like about
Stockholm is it has three equalqualities that you need in AI
stuff.
Well there is a very muchcentralized decision making.
Very decentralizedaccountability.
That means people are,individually very innovative and
they are thinking about theproblems and solutions.
(03:27):
At the same time, there is avery collaborative culture.
So collaboration is veryimportant here in Stockholm.
And we see people collaboratingvery openly in a very trusted
environment.
And that kind of builds uponeach other's works, each other
stores, and help us come up withvery holistic solutions.
Andreas Welsch (03:44):
So great job answering on the fly.
Thank you so much.
And I already heard you saysomething about transformation,
right?
And that it's a significanteffort.
So let's take a look at thefirst question here.
How do you define digitaltransformation?
Maybe Kieran, can you help mewith this?
Kieran Gilmurray (04:04):
This is got a big one.
Let me talk about three Ps: people, process, and product. (04:05):
undefined
So if we're talking aboutdigital transformation, we're
talking about a mindset change.
It's not a case of throwingsomething to the IT department
saying transform it because it'stechnology.
Everyone in the organizationneeds to think digital, behave
digitally, and act digitally.
We say act digitally.
What do we mean?
(04:26):
No longer analog.
All the things that they dealwith are digital.
The tooling, the platform, andagain, the product as well.
So we're, once, we were dealingwith a lot of physical products,
now we're dealing with a lot ofdigital products.
So those three things areprimarily then you're trying to
get all those things together.
So when we're talking aboutdigital transformation, we're
talking about Change culturechanging people as well as the
(04:49):
technology.
A lot of people focus on thetech side, Andreas, but people
transform businesses, nottechnology.
They used to just use reallygreat technology to augment what
they do.
So once everybody's got theright mindset, they're gonna do
it digitally and that mattersfrom everyone from the post room
right the way up to the chairmanof the company.
And I'm a tremendous fan ofsaying an executive team needs
(05:09):
to understand technology becausemost business, strateg.
These days are technologyenabled.
It's a requirement just likeunderstanding finances these
days.
Once you've built anorganization to act digitally
and think digitally, yourproducts tend to end up being
digital products or there's apile of AI or data surrounding
those products allowing you todeliver new value very quickly.
(05:31):
An example, Insurance brokersyears ago used to pick up a
telephone, ring them.
You would get a piece ofinsurance sent out in the post
numbers of days later.
Now you're able to log intoonline websites that work out
who you are.
Work out your preferences.
Work out next.
Next best action.
Next best offer.
Product upsell, cross-sellcredit ratings, debt ratings,
renewal retention rates.
(05:52):
Chances are when that goesthrough the system, you're
ending up with a digitalcertificate.
because the computing systemsinside of the company are all
based on a digital platform witha digital front, and there is
very little manual paperwork.
Everything's really instant.
And when it comes to making aclaim, take a picture with your
phone, use AI to describe theaccident scene, work out the
value order, the breakdownvehicle, order, the garage
(06:14):
that's going to pick up the car,give you a cost estimate, give
you another vehicle or a hotel,depending on a far you are from
home.
All done automated without thenecessity to have someone there.
But at the same time, all ofthis tech and AI and everything
else around it.
You've digitally transformedyour firm should be sensitive to
you as an individual, andtherefore it should connect you
(06:36):
with an amazing human whounderstands intimately your
needs because they've profiledyou, done used whatever big data
sets that exist on the marketand are able to answer the
question where humans addtremendous value and to be able
to do, again, very digitally.
Camera, face, phone, tech,WhatsApp, you name it, whatever
channel door or whatever way youwant to.
(06:57):
They're there to do it, and allthat data is connected together.
So there's one customer view,everybody's aware of it, new as
a customer, feel like you'vebeen treated as an individual,
but chances are you're one ofmany customers.
So people, product place orplatform, sorry, mindset change.
Method change, and the actual.
Digital product that you've gotat the end of the day, that's
(07:17):
changed too.
And then there'll be a whole setof digital metrics around that
to measure your performance.
And all the time you'reattempting to remove attrition
from the process usingtechnology and constantly
driving forward.
Otherwise you run the risk ofbeing outta business.
That's a process.
Someone once called DigitalDarwinism and I love that
phraseology.
Andreas Welsch (07:38):
So great points all around.
Thanks, Kieran.
The part that really resonateswith me is that about people.
Because when I typically thinkof transformation, I think of
changing the employeeexperience, changing something
for them.
But I think you made anexcellent point that it's not
only just about the employee,but actually also the customer.
(07:59):
And spinning that further Ithink that's where Somil, you
and the algorithmic decisionmaking and your focus on that
come in.
So maybe let's take a look atquestion number two.
And maybe Somil, that issomething that, that you can
help me with.
The question is, what docompanies need to do to monetize
(08:23):
AI?
So over to you, Somil.
Somil Gupta (08:27):
So one of the things, when you look at
monetization is I see it's a lotabout mindset.
And one of the things which I'ma bit concerned is more, a lot
of companies, they're working ina very project centric mindset.
That means they're focusing alot on activities and not on
outcomes.
And I like to think of, when Ithink about modernization and
(08:48):
look at solutions, I like tothink of them as like a four
concentric circles.
So the innermost circles yourdata model, and that's.
Everything, the value that youhave that is called your data,
and then you go one circleoutside and then you have the
decision intelligence and youhave your decision.
And that's where you use yourai, you use your, and you use
(09:08):
different techniques, make senseof that data and kind of convert
that into insights and peopletend to stop there.
But there are two additionalcircles.
The moment you go outside thisdecision intelligence, you get
into the commercial model.
And commercial model is reallyabout relationships.
It's about the processes.
Why are we a business?
What kind of relationship are weenforcing?
(09:29):
Or what kind of relationship arewe enabling?
And then there's finally theexternal model, which is the
business model inside which allof this happens, right?
And that is more about how do wemake money?
How do we lead?
If you want to go from lookingat AI as project that you
implement, or AI as a model, ifyou wanna go towards
modernization, we have to followthis three step process.
(09:52):
And it's very simple.
I call it PCM (09:53):
productize, commercialize, and monetize.
So productization is reallyabout packaging data and AI
models in such a way where it'snot only valuable, but it's also
easily consumable.
That means you have to go thatextra mile towards your users,
helping them derive value.
And I actually believe thathelping the end user derive
(10:15):
value from your product isreally the responsibility of the
product manager.
It's not the consumer'sresponsibility.
Many people say they have tocome meet halfway, but I think
we have to go all the way.
Commercialization that is reallyone part is scaling and
initializing it.
Other part is also to build inour ecosystem.
How are you going to deliverthis to your customer?
How you're going to operate it?
(10:35):
How you reorchestrate it?
How you deliver services?
That whole operating modelprinciple comes into
commercialization.
And finally, modernization iswhere you get paid for all the
hard work that you've done.
That goes to your contracts, toyour pricing.
And how do you really createvalue for the end user, but also
for yourself for being thisentrepreneurial venture.
(10:57):
So I think that's whatmonetizing should look like.
And we need to really moveahead, move away from this
project mindset.
A lot of companies are trying toget us work management around
data and AI.
It's not really working out andget towards the product mindset,
which is more focused onoutcomes and more focused on
kind of incremental valuecreation.
Andreas Welsch (11:14):
Thanks, Somil.
I really like that part that youmentioned, to think more broadly
and not to approach it just likeany other IT project.
I think that's really key.
So maybe Kieran for you.
So maybe you can build on that.
Are you seeing a lot ofcustomers already doing that?
And a lot of companies like inthe insurance example, really
(11:38):
doing that, monetizing that?
Kieran Gilmurray (11:42):
Yeah, you do.
Yeah, done right?
You do you know what I mean?
Because ultimately, why are wein business?
Of course, this meal says it'snot about just doing activities,
it's about actually creatingsomething that's of value that
someone is willing to pay moneyfor.
So you can use AI in lots ofways.
The example we gave a momentago, insurance.
Yes, of course.
Next best action, cross sale,up, sale, renewal, retention,
(12:02):
figures, whatever else.
But I've seen some.
Great companies having been setup literally to collect data,
and then they're using the datato sell the data on for
different purposes.
So Facebook's an amazing exampleof that.
At TikTok, you name it, allthese companies are set up to
collect vast amounts of data,and then they use that data to
allow other companies to target,or they sell services like
(12:25):
LinkedIn, to do targetingmarketing, you name it.
You can use the data that you'vegot for a tremendous amount, in
a tremendous amount of differentways to actually derive value.
The other bit as well is we'retalking about, that's external
value for me.
We're looking at the internalvalue as well.
So if I start to get IoT devicesand you're start seeing a lot of
shipping in companies to be alot more.
(12:46):
Careful around what they'redoing, and this can be anything
from literally logistics,shipping from one country to
another, and then use theanalytics to work out when
should I travel?
How long should I travel, howquick should I travel?
You'll see the same thing withairlines or cargo ships or
whatever else, right?
The way to building digitaltwins, for example, if I've got
an oil field sitting somewherein the North Sea or in wherever
it is, you start to seecompanies using AI to build
(13:09):
digital twins of those, tooperate them in the most
efficient manner possible.
You also see them using AI to dopredictive maintenance on
equipment because every hourthat those particular facilities
are down, it's not, hundreds ofdollars.
It's potentially hundreds ofmillions of dollars with these
things are not actuallyoperating.
You see the real value of thatnow in the world where energy
(13:29):
and production costs and oilcosts and whatever else are
going through the roof.
They're really smart.
Companies are looking at everysingle part of their value
chain.
Both the customer end, their ownperformance end and everything
in between.
And then they're using otheranalytics in tremendous ways
around the people analytics.
Looking at how likely people areto stay looking at analytics as
(13:50):
to who they should hire.
Looking at analytics to work outhow do you put teams together to
work in the best possible, mostefficient way.
And everything else in between.
So AI analytics are amazinglypowerful when you use them
across the entire value chain.
But you need to be clear, assomeone says in the very first
instance, what are the outcomesyou're actually trying to get?
(14:13):
And some of those might be thirdparty outcomes, deliberately
like Facebook.
Some of those might bedeliberate.
Building a digital twin andworking off that to drive
efficiency.
And some of them could actuallybe what I describe as making
money off the exhaust fumes ofcompanies activities.
So I've seen some companies thatare set up to do quoting, they
can't actually quote for aparticular piece of business.
(14:33):
And what they've done is use theAI or the data that they've
collected to sell on to othercompanies, and they're making
money outta that AI as well.
A clever example is a, again, alarge company in the consumer
space, they get thousands ofpeople and companies giving them
their very latest details everyday to get.
Quoted on a particular type ofbusiness I can't name as private
(14:55):
company.
And then what they actually dois sell that company on to other
insurance companies, othertargeting companies, other
companies that validate thatthese companies exist with the
latest contact details and thedirectors and everything else to
do credit checks and so there'sreally clever things happening
across the company's valuechain.
And then they're taking thatdata, selling it on to allow
(15:17):
other companies to dointeresting and different
things.
But always with a business iconin mind.
Andreas Welsch (15:23):
So here, and I think you're making a key point
here, right?
That it's really about thebusiness outcome and the
business value and making all ofthat measurable.
Because a lot of times I alsohear about, we're doing
automation and it'll lead to abetter employee experience.
And I think that's important andgood and fair.
Of course we want our employeesto feel.
(15:43):
Good about their work and behappy and enjoy coming to work.
But that's certainly not enough,right?
It needs to be more, and itneeds to be measurable.
So I really like that componentthat you are highlighting that
here.
And so maybe question for you,Somil.
Kieran already talked aboutinsurance is one example, but
what types of data are youseeing companies monetize with?;
Somil Gupta (16:10):
So as the field itself is still evolving for
reason because a lot ofcompanies are still trying to
get their head around the dataquality, and for me, data
quality is always fit forpurpose and then know depends a
lot on what you want to do.
But lot of companies have comeupon themselves to define the
pristine.
Got like data, whatever theywanna have.
(16:33):
It's more world of fantasy.
IoT data is one thing that wesee most of the companies they
want to now monetize because youget like real information about
what's happening on the ground.
Logistics data, definitely verymuch in use.
I think financial data thatcompanies are producing,
(16:53):
transactional data.
It's not so much for me aboutwhat data, but it's like what
purpose they're driving.
The data sources can come fromboth internally, externally,
from devices, but it's moreabout what companies need to
figure out is how are wedifferentiating ourself?
(17:14):
How are we positioning ourself?
How are we really creating valuein this thing?
So IoT is, for example, one ofthe things that people have
interest in.
But I think there's a long wayto go to really integrate that
data in other source of data tocreate something tangible.
Andreas Welsch (17:28):
I see.
When we talk about data, accessto data, different types of
data, monetizing it forimproving the customer
experience, improving theemployee experience.
I think that's a key point here.
So maybe let's take a look atquestion number three.
(17:49):
And that is, what is the role ofethics in this?
Maybe that's something you canhelp me with.
Kieran, can you take a stab atthat?
Kieran Gilmurray (17:59):
Yeah it's a massive topic and has probably
come to the fore in the pressaround that company mentioned
earlier on, Facebook and we hadCambridge Analytica who took the
data and used that to what youmight describe, profile people
and then try and persuade peopleto do particular things.
An example of.
Was it the best use of datapossibly for the company?
(18:22):
Was it the best use of data forpolitical parties and people
themselves?
Debatable, probably not.
It's all new technologies.
This is an interesting onebecause if you're talking about
IT, programming in general, nowwe're worried about security and
people, company security andfinancial loss ethics has been
one of those things whereeverybody got really excited
about the power of and theyprobably got so excited that
(18:44):
they concentrated on doing theactual work.
But when they worked all thesethings through, they discovered
that there was a genuine peopleimpact.
Let me go back to insurance andgive you an example.
A lot of companies run retentionanalytics.
So what they do is they build amodel.
They work out which customersare likely to stay with them and
which customers are not likelyto.
(19:05):
Very often what they do is theyactually price the insurance
according to that.
So if you're highly likely tostay, you will not get a
discount from a broker orinsurance company.
Chances are you'll actually geta price rise if you're highly
likely to leave.
They will probably offer you adiscount to stay.
Statistically speaking, becauseI've been in the industry.
(19:25):
The older generation, and again,whatever age you wanna put that,
50, 67, they tend not to lookaround, tend to trust at times,
tended not to be as digitallyliterate.
So what you're actually doing istargeting people who had.
Very little propensity to moveand you are constantly raising
the prices, whereas someone in adifferent cohort would've left
(19:47):
and they were getting adifferent price for exactly the
same product, not exactlyethically sound, and probably
against numbers of insuranceregulations at the same time.
You do see that as well in.
Referencing and credit agencies.
If you are someone on a lowerincome and have a per credit
history, then when it comes toaccessing credit and buying
(20:09):
something like a car or afinancial product or something
else, a house of mortgage tocover that risk company's
actually gonna charge you moremoney.
Now, again, you know at.
May be correct from a companypoint of view, but how ethical
is it that you or I who arecoming as two consumers who are
probably the same but Mightlyhave slightly different credit
history, are getting twodifferent prices.
(20:31):
So anywhere from potentialmanipulation of people,
manipulation of pricing,treating people differently, who
are getting the same product.
All these are is are technicallyand have risen.
Poor ethical practices and pooroutcomes of using analytics that
resulted in some harm or othertwo consumers.
(20:51):
From a company point of viewthough, just remember this, if
we're talking purely financial,using analytics to make those
decisions to derive the mostprofit that you possibly can.
Is something that most companieswant to strive for.
Therefore, we end up with anethical debate.
How ethical is it for a companythat's set up to make profit, to
use every tool at its advantage?
(21:12):
And again, if you have superiortechnology, you're gonna do
better than a company thatdoesn't.
But data analytics is one ofthose things that's really risen
to the fore run, this particulartopic, it's really powerful
where machines can crunch largevolumes of data that companies
have into Samir's Point.
They can also now buy other setsof data to allow them to make
more micro decisions.
(21:33):
They can be more persuasiveusing data.
If you combine that data withbehavioral psychology NLP
neurolinguistic programming, youcan start to tangibly
manipulate, and I use that word,both positively and negatively,
how people will respond to yourproduct and your price and
everything else in between.
I personally think.
Like data companies need tothink through all of the
(21:56):
operations that they're doing.
Companies who use data need tothink all the way through to the
end of their operations.
The need to focus on consumersto make sure they're providing
great value and not manipulatingthem.
Can we trust companies to dothat?
Question mark.
I don't think every company inthe world is a horrible,
unethical group.
You might assume it the way thepress treats companies who deal
(22:16):
with ai.
I do think there's companies outthere that do forget their
responsibility, their biggerresponsibility to society and
individuals to treat them fairlyand equitably.
Do we need a regulator?
Question mark.
The medical industry, forexample, as a regulator to make
sure that people build medicalor medicines that don't harm
people.
Is it arguable that becausedata's become so prevalent and
(22:39):
so important in society, that weneed a regulator for a lot to
allow companies to behavehonorably and equitably
possibly, and probably.
Andreas Welsch (22:48):
Thanks.
So ethics obviously play plays akey role.
I know there is a lot of debatehappening between experts and
also on a government level, likeat the E.U.
with the AI Act or even overhere in, in the U.S.
But maybe question for you,building on that Somil what do
you see in your work?
(23:10):
How early should companiesreally think about ethics when
they do work around AI?
And who should actually own thistopic of AI ethics?
Somil Gupta (23:20):
Yeah, I focused a lot on the ethical
commercialization, ethicalmodernization.
The only kind of thing is that Ithink the ethics discussion
should not be an afterthought.
Ethics has to be built into thedesign of the product.
Before we even start talkingabout a product, we start
building the value framework.
And in that value framework, wehave to start embedding ethical
(23:41):
constraints and and ethicalpolicies.
The reason why this is importantis if you're not, put that in
the initial part of framing ofthe problem, both technical
framing and commercial framing,then what happens is that even
while you get your model up andrunning, it's totally too late
to look after ethics after themodel is being built.
(24:03):
And then we will be like tothink that our data scientists
are smart enough to figure outhow the model is responding, and
that is completely not true.
Once our model is deployed, ittends to have its own life and
how it's making prediction.
Of course, you could figure itout later on if you do a
diagnostics, but then it'salready too late.
(24:23):
Denial of service, infringing ofsomebody's rights manipulation
social engineering, all thoseare real threats now.
You cannot wait late enough.
You can't put early enough.
I can't stress upon this.
The framework that we create,the value framework.
In that value framework that wecreated the first time, that is
(24:43):
where we have to start lookingat ethics.
Because you start looking at whoare the people who are going to
be most vulnerable?
I think you give an example ofthe people who are elderly, who
might be impacted the most, butwho are the least who also the
most vulnerable.
We also saw some examples of gigeconomy, like drivers within
Uber who get impacted with analgorithm, on which they have.
(25:05):
Uber does not have any say onthis.
They can't figure out whetherthe decision was taken or was
right or wrong.
You have to understand that thedecision intelligence works
within the commercial framework.
You cannot control the decisionfrom point to point.
But we can definitely controlthe commercial framework which
we operate.
And we have to realize thatthese companies, they have
massive amounts of data,thousands of data points about
(25:25):
us.
They have a lot of power.
In what they can do.
So then really comes down tothis ethical design of
offerings, ethical pricing ofofferings.
Even though a risk like dynamicpricing.
So when we are doing dynamicpricing, we should not infringe
upon somebody's rights.
We should not discriminateagainst people.
So that's the kind of thingwhich is a trade off between
(25:48):
doing what is right versus doingwhat is easy.
And I hate to say that, but alot of companies, they're not
really inputting a thoughtaround these topics.
So all these ethics andeverything, governance, they
come very late in the cycle bythen, it's usually too late to
do anything about it.
Andreas Welsch (26:05):
Fantastic.
So when it comes to ethics,start early and make sure it's
pervasive throughout theproject.
That's awesome, team! We've madeit.
Thank you so much for helping meplay"What's the BUZZ?" today?
Kieran Gilmurray (26:18):
Hurray!
Andreas Welsch (26:20):
So let me quickly summarize.
First of all, if you think aboutdigital transformation, keep the
three Ps in mind (26:26):
people, place, and platform.
But really people should be thefocus and the center of your
initiatives, whether it's youremployees or your customers,
because the transformationcertainly impacts them in
different ways.
Secondly, if you're thinkingabout monetization of AI, really
(26:47):
don't treat it like any other ITproject, because it's not.
The different layers of it inand AI should really be core.
And as you're doing that, youget to number three, make sure
that ethics is a part of it fromthe very beginning, all the way
to the end of that project.
So we're getting close to theend of the show.
Somil Gupta (27:06):
Thanks, Andreas.
Today we have a special episode for you.
Now, I've been playing one of myfavorite retro game:"What's the
BUZZ?" But I'm stuck.
And that's why I've asked a fewexperts to help me"win with AI",
Somil Gupta and KieranGilmurray.
Thanks for joining.
Kieran Gilmurray (00:17):
Great to be here.
I've been in business technologyfor about the last 27 years.
I've described both because Ilove technology, but love what
it can do for a business.
I've sat in intelligentautomation roles.
I've ran AI teams for 13, 14years.
I've ran businesses, large,small, international, you name
it, I've been there, done that.
Hopefully today I can add somevalue.
(00:37):
Recently I've written a book onthe A to Z to digital
transformation.
I'm going to use some of theanswers in that book, hopefully
in today's session as well.
Delighted to be here.
Somil Gupta (00:46):
Thank you.
My name is Somil Gupta.
I'm the founder of AlgorithmicScale.
It's a company based in Sweden,and we focus on building the
monetization and commercialstack for AI based business
models.
What it means is we help ourclients build production
services.
We help them build digitaloperating models for delivering
(01:07):
the value that they create fromfrom AI.
But also help them with valuecare, which is helping them with
pricing strategies, helping themwith different modernization
strategies.
And this is something which kindof very exciting.
So I am very crazy aboutalgorithmic operating models and
algorithmic business models.
And now this company issomething that will intended to
(01:30):
make those visions into reality.
Andreas Welsch (01:34):
Awesome.
Thank you so much.
If you're just joining thisstream, drop a comment in the
chat, what tricks you arelooking for.
No cheating allowed by the way.
So what do you say?
Should we start playing?
Kieran Gilmurray (01:48):
Oh, yes,
Andreas Welsch (01:50):
Okay, so this one is a warmup.
When I hit the buzzer, you'llsee a sentence.
I need your help.
So can you answer with the firstthing that comes to mind in why,
in your own words?
So you only have 60 seconds foryour answer.
And for those of you watching uslive, please drop yours in the
(02:12):
chat as well.
Team, are you ready for What'sthe BUZZ?
If AI were city, which would itbe?
60 seconds.
Go.
Kieran Gilmurray (02:25):
Oh wow.
Let me say Boston for a coupleof reasons.
What an amazing practical city.
There is 1,001 amazing things.
There are all wrapped up in areasonably small package, and
when everybody goes to Boston,they all experience something
different.
Everybody thinks of it as adifferent place.
So I think it was a us Irishcity.
(02:45):
I'm sure a lot of other peopleremembered from all sorts of
movies and different things, buta really practical, thriving
city where a lot of tech exists,where a lot of people are making
an impact and there's still lotsof potential for things crazy
great things to happen.
Somil Gupta (03:01):
All right.
I would say Stockholm.
And reason being, I think one ofthe things which I like about
Stockholm is it has three equalqualities that you need in AI
stuff.
Well there is a very muchcentralized decision making.
Very decentralizedaccountability.
That means people are,individually very innovative and
they are thinking about theproblems and solutions.
(03:27):
At the same time, there is avery collaborative culture.
So collaboration is veryimportant here in Stockholm.
And we see people collaboratingvery openly in a very trusted
environment.
And that kind of builds uponeach other's works, each other
stores, and help us come up withvery holistic solutions.
Andreas Welsch (03:44):
So great job answering on the fly.
Thank you so much.
And I already heard you saysomething about transformation,
right?
And that it's a significanteffort.
So let's take a look at thefirst question here.
How do you define digitaltransformation?
Maybe Kieran, can you help mewith this?
Kieran Gilmurray (04:04):
This is got a big one.
Let me talk about three Ps: people, process, and product. (04:05):
undefined
So if we're talking aboutdigital transformation, we're
talking about a mindset change.
It's not a case of throwingsomething to the IT department
saying transform it because it'stechnology.
Everyone in the organizationneeds to think digital, behave
digitally, and act digitally.
We say act digitally.
What do we mean?
(04:26):
No longer analog.
All the things that they dealwith are digital.
The tooling, the platform, andagain, the product as well.
So we're, once, we were dealingwith a lot of physical products,
now we're dealing with a lot ofdigital products.
So those three things areprimarily then you're trying to
get all those things together.
So when we're talking aboutdigital transformation, we're
talking about Change culturechanging people as well as the
(04:49):
technology.
A lot of people focus on thetech side, Andreas, but people
transform businesses, nottechnology.
They used to just use reallygreat technology to augment what
they do.
So once everybody's got theright mindset, they're gonna do
it digitally and that mattersfrom everyone from the post room
right the way up to the chairmanof the company.
And I'm a tremendous fan ofsaying an executive team needs
(05:09):
to understand technology becausemost business, strateg.
These days are technologyenabled.
It's a requirement just likeunderstanding finances these
days.
Once you've built anorganization to act digitally
and think digitally, yourproducts tend to end up being
digital products or there's apile of AI or data surrounding
those products allowing you todeliver new value very quickly.
(05:31):
An example, Insurance brokersyears ago used to pick up a
telephone, ring them.
You would get a piece ofinsurance sent out in the post
numbers of days later.
Now you're able to log intoonline websites that work out
who you are.
Work out your preferences.
Work out next.
Next best action.
Next best offer.
Product upsell, cross-sellcredit ratings, debt ratings,
renewal retention rates.
(05:52):
Chances are when that goesthrough the system, you're
ending up with a digitalcertificate.
because the computing systemsinside of the company are all
based on a digital platform witha digital front, and there is
very little manual paperwork.
Everything's really instant.
And when it comes to making aclaim, take a picture with your
phone, use AI to describe theaccident scene, work out the
value order, the breakdownvehicle, order, the garage
(06:14):
that's going to pick up the car,give you a cost estimate, give
you another vehicle or a hotel,depending on a far you are from
home.
All done automated without thenecessity to have someone there.
But at the same time, all ofthis tech and AI and everything
else around it.
You've digitally transformedyour firm should be sensitive to
you as an individual, andtherefore it should connect you
(06:36):
with an amazing human whounderstands intimately your
needs because they've profiledyou, done used whatever big data
sets that exist on the marketand are able to answer the
question where humans addtremendous value and to be able
to do, again, very digitally.
Camera, face, phone, tech,WhatsApp, you name it, whatever
channel door or whatever way youwant to.
(06:57):
They're there to do it, and allthat data is connected together.
So there's one customer view,everybody's aware of it, new as
a customer, feel like you'vebeen treated as an individual,
but chances are you're one ofmany customers.
So people, product place orplatform, sorry, mindset change.
Method change, and the actual.
Digital product that you've gotat the end of the day, that's
(07:17):
changed too.
And then there'll be a whole setof digital metrics around that
to measure your performance.
And all the time you'reattempting to remove attrition
from the process usingtechnology and constantly
driving forward.
Otherwise you run the risk ofbeing outta business.
That's a process.
Someone once called DigitalDarwinism and I love that
phraseology.
Andreas Welsch (07:38):
So great points all around.
Thanks, Kieran.
The part that really resonateswith me is that about people.
Because when I typically thinkof transformation, I think of
changing the employeeexperience, changing something
for them.
But I think you made anexcellent point that it's not
only just about the employee,but actually also the customer.
(07:59):
And spinning that further Ithink that's where Somil, you
and the algorithmic decisionmaking and your focus on that
come in.
So maybe let's take a look atquestion number two.
And maybe Somil, that issomething that, that you can
help me with.
The question is, what docompanies need to do to monetize
(08:23):
AI?
So over to you, Somil.
Somil Gupta (08:27):
So one of the things, when you look at
monetization is I see it's a lotabout mindset.
And one of the things which I'ma bit concerned is more, a lot
of companies, they're working ina very project centric mindset.
That means they're focusing alot on activities and not on
outcomes.
And I like to think of, when Ithink about modernization and
(08:48):
look at solutions, I like tothink of them as like a four
concentric circles.
So the innermost circles yourdata model, and that's.
Everything, the value that youhave that is called your data,
and then you go one circleoutside and then you have the
decision intelligence and youhave your decision.
And that's where you use yourai, you use your, and you use
(09:08):
different techniques, make senseof that data and kind of convert
that into insights and peopletend to stop there.
But there are two additionalcircles.
The moment you go outside thisdecision intelligence, you get
into the commercial model.
And commercial model is reallyabout relationships.
It's about the processes.
Why are we a business?
What kind of relationship are weenforcing?
(09:29):
Or what kind of relationship arewe enabling?
And then there's finally theexternal model, which is the
business model inside which allof this happens, right?
And that is more about how do wemake money?
How do we lead?
If you want to go from lookingat AI as project that you
implement, or AI as a model, ifyou wanna go towards
modernization, we have to followthis three step process.
(09:52):
And it's very simple.
I call it PCM (09:53):
productize, commercialize, and monetize.
So productization is reallyabout packaging data and AI
models in such a way where it'snot only valuable, but it's also
easily consumable.
That means you have to go thatextra mile towards your users,
helping them derive value.
And I actually believe thathelping the end user derive
(10:15):
value from your product isreally the responsibility of the
product manager.
It's not the consumer'sresponsibility.
Many people say they have tocome meet halfway, but I think
we have to go all the way.
Commercialization that is reallyone part is scaling and
initializing it.
Other part is also to build inour ecosystem.
How are you going to deliverthis to your customer?
How you're going to operate it?
(10:35):
How you reorchestrate it?
How you deliver services?
That whole operating modelprinciple comes into
commercialization.
And finally, modernization iswhere you get paid for all the
hard work that you've done.
That goes to your contracts, toyour pricing.
And how do you really createvalue for the end user, but also
for yourself for being thisentrepreneurial venture.
(10:57):
So I think that's whatmonetizing should look like.
And we need to really moveahead, move away from this
project mindset.
A lot of companies are trying toget us work management around
data and AI.
It's not really working out andget towards the product mindset,
which is more focused onoutcomes and more focused on
kind of incremental valuecreation.
Andreas Welsch (11:14):
Thanks, Somil.
I really like that part that youmentioned, to think more broadly
and not to approach it just likeany other IT project.
I think that's really key.
So maybe Kieran for you.
So maybe you can build on that.
Are you seeing a lot ofcustomers already doing that?
And a lot of companies like inthe insurance example, really
(11:38):
doing that, monetizing that?
Kieran Gilmurray (11:42):
Yeah, you do.
Yeah, done right?
You do you know what I mean?
Because ultimately, why are wein business?
Of course, this meal says it'snot about just doing activities,
it's about actually creatingsomething that's of value that
someone is willing to pay moneyfor.
So you can use AI in lots ofways.
The example we gave a momentago, insurance.
Yes, of course.
Next best action, cross sale,up, sale, renewal, retention,
(12:02):
figures, whatever else.
But I've seen some.
Great companies having been setup literally to collect data,
and then they're using the datato sell the data on for
different purposes.
So Facebook's an amazing exampleof that.
At TikTok, you name it, allthese companies are set up to
collect vast amounts of data,and then they use that data to
allow other companies to target,or they sell services like
(12:25):
LinkedIn, to do targetingmarketing, you name it.
You can use the data that you'vegot for a tremendous amount, in
a tremendous amount of differentways to actually derive value.
The other bit as well is we'retalking about, that's external
value for me.
We're looking at the internalvalue as well.
So if I start to get IoT devicesand you're start seeing a lot of
shipping in companies to be alot more.
(12:46):
Careful around what they'redoing, and this can be anything
from literally logistics,shipping from one country to
another, and then use theanalytics to work out when
should I travel?
How long should I travel, howquick should I travel?
You'll see the same thing withairlines or cargo ships or
whatever else, right?
The way to building digitaltwins, for example, if I've got
an oil field sitting somewherein the North Sea or in wherever
it is, you start to seecompanies using AI to build
(13:09):
digital twins of those, tooperate them in the most
efficient manner possible.
You also see them using AI to dopredictive maintenance on
equipment because every hourthat those particular facilities
are down, it's not, hundreds ofdollars.
It's potentially hundreds ofmillions of dollars with these
things are not actuallyoperating.
You see the real value of thatnow in the world where energy
(13:29):
and production costs and oilcosts and whatever else are
going through the roof.
They're really smart.
Companies are looking at everysingle part of their value
chain.
Both the customer end, their ownperformance end and everything
in between.
And then they're using otheranalytics in tremendous ways
around the people analytics.
Looking at how likely people areto stay looking at analytics as
(13:50):
to who they should hire.
Looking at analytics to work outhow do you put teams together to
work in the best possible, mostefficient way.
And everything else in between.
So AI analytics are amazinglypowerful when you use them
across the entire value chain.
But you need to be clear, assomeone says in the very first
instance, what are the outcomesyou're actually trying to get?
(14:13):
And some of those might be thirdparty outcomes, deliberately
like Facebook.
Some of those might bedeliberate.
Building a digital twin andworking off that to drive
efficiency.
And some of them could actuallybe what I describe as making
money off the exhaust fumes ofcompanies activities.
So I've seen some companies thatare set up to do quoting, they
can't actually quote for aparticular piece of business.
(14:33):
And what they've done is use theAI or the data that they've
collected to sell on to othercompanies, and they're making
money outta that AI as well.
A clever example is a, again, alarge company in the consumer
space, they get thousands ofpeople and companies giving them
their very latest details everyday to get.
Quoted on a particular type ofbusiness I can't name as private
(14:55):
company.
And then what they actually dois sell that company on to other
insurance companies, othertargeting companies, other
companies that validate thatthese companies exist with the
latest contact details and thedirectors and everything else to
do credit checks and so there'sreally clever things happening
across the company's valuechain.
And then they're taking thatdata, selling it on to allow
(15:17):
other companies to dointeresting and different
things.
But always with a business iconin mind.
Andreas Welsch (15:23):
So here, and I think you're making a key point
here, right?
That it's really about thebusiness outcome and the
business value and making all ofthat measurable.
Because a lot of times I alsohear about, we're doing
automation and it'll lead to abetter employee experience.
And I think that's important andgood and fair.
Of course we want our employeesto feel.
(15:43):
Good about their work and behappy and enjoy coming to work.
But that's certainly not enough,right?
It needs to be more, and itneeds to be measurable.
So I really like that componentthat you are highlighting that
here.
And so maybe question for you,Somil.
Kieran already talked aboutinsurance is one example, but
what types of data are youseeing companies monetize with?;
Somil Gupta (16:10):
So as the field itself is still evolving for
reason because a lot ofcompanies are still trying to
get their head around the dataquality, and for me, data
quality is always fit forpurpose and then know depends a
lot on what you want to do.
But lot of companies have comeupon themselves to define the
pristine.
Got like data, whatever theywanna have.
(16:33):
It's more world of fantasy.
IoT data is one thing that wesee most of the companies they
want to now monetize because youget like real information about
what's happening on the ground.
Logistics data, definitely verymuch in use.
I think financial data thatcompanies are producing,
(16:53):
transactional data.
It's not so much for me aboutwhat data, but it's like what
purpose they're driving.
The data sources can come fromboth internally, externally,
from devices, but it's moreabout what companies need to
figure out is how are wedifferentiating ourself?
(17:14):
How are we positioning ourself?
How are we really creating valuein this thing?
So IoT is, for example, one ofthe things that people have
interest in.
But I think there's a long wayto go to really integrate that
data in other source of data tocreate something tangible.
Andreas Welsch (17:28):
I see.
When we talk about data, accessto data, different types of
data, monetizing it forimproving the customer
experience, improving theemployee experience.
I think that's a key point here.
So maybe let's take a look atquestion number three.
(17:49):
And that is, what is the role ofethics in this?
Maybe that's something you canhelp me with.
Kieran, can you take a stab atthat?
Kieran Gilmurray (17:59):
Yeah it's a massive topic and has probably
come to the fore in the pressaround that company mentioned
earlier on, Facebook and we hadCambridge Analytica who took the
data and used that to what youmight describe, profile people
and then try and persuade peopleto do particular things.
An example of.
Was it the best use of datapossibly for the company?
(18:22):
Was it the best use of data forpolitical parties and people
themselves?
Debatable, probably not.
It's all new technologies.
This is an interesting onebecause if you're talking about
IT, programming in general, nowwe're worried about security and
people, company security andfinancial loss ethics has been
one of those things whereeverybody got really excited
about the power of and theyprobably got so excited that
(18:44):
they concentrated on doing theactual work.
But when they worked all thesethings through, they discovered
that there was a genuine peopleimpact.
Let me go back to insurance andgive you an example.
A lot of companies run retentionanalytics.
So what they do is they build amodel.
They work out which customersare likely to stay with them and
which customers are not likelyto.
(19:05):
Very often what they do is theyactually price the insurance
according to that.
So if you're highly likely tostay, you will not get a
discount from a broker orinsurance company.
Chances are you'll actually geta price rise if you're highly
likely to leave.
They will probably offer you adiscount to stay.
Statistically speaking, becauseI've been in the industry.
(19:25):
The older generation, and again,whatever age you wanna put that,
50, 67, they tend not to lookaround, tend to trust at times,
tended not to be as digitallyliterate.
So what you're actually doing istargeting people who had.
Very little propensity to moveand you are constantly raising
the prices, whereas someone in adifferent cohort would've left
(19:47):
and they were getting adifferent price for exactly the
same product, not exactlyethically sound, and probably
against numbers of insuranceregulations at the same time.
You do see that as well in.
Referencing and credit agencies.
If you are someone on a lowerincome and have a per credit
history, then when it comes toaccessing credit and buying
(20:09):
something like a car or afinancial product or something
else, a house of mortgage tocover that risk company's
actually gonna charge you moremoney.
Now, again, you know at.
May be correct from a companypoint of view, but how ethical
is it that you or I who arecoming as two consumers who are
probably the same but Mightlyhave slightly different credit
history, are getting twodifferent prices.
(20:31):
So anywhere from potentialmanipulation of people,
manipulation of pricing,treating people differently, who
are getting the same product.
All these are is are technicallyand have risen.
Poor ethical practices and pooroutcomes of using analytics that
resulted in some harm or othertwo consumers.
(20:51):
From a company point of viewthough, just remember this, if
we're talking purely financial,using analytics to make those
decisions to derive the mostprofit that you possibly can.
Is something that most companieswant to strive for.
Therefore, we end up with anethical debate.
How ethical is it for a companythat's set up to make profit, to
use every tool at its advantage?
(21:12):
And again, if you have superiortechnology, you're gonna do
better than a company thatdoesn't.
But data analytics is one ofthose things that's really risen
to the fore run, this particulartopic, it's really powerful
where machines can crunch largevolumes of data that companies
have into Samir's Point.
They can also now buy other setsof data to allow them to make
more micro decisions.
(21:33):
They can be more persuasiveusing data.
If you combine that data withbehavioral psychology NLP
neurolinguistic programming, youcan start to tangibly
manipulate, and I use that word,both positively and negatively,
how people will respond to yourproduct and your price and
everything else in between.
I personally think.
Like data companies need tothink through all of the
(21:56):
operations that they're doing.
Companies who use data need tothink all the way through to the
end of their operations.
The need to focus on consumersto make sure they're providing
great value and not manipulatingthem.
Can we trust companies to dothat?
Question mark.
I don't think every company inthe world is a horrible,
unethical group.
You might assume it the way thepress treats companies who deal
(22:16):
with ai.
I do think there's companies outthere that do forget their
responsibility, their biggerresponsibility to society and
individuals to treat them fairlyand equitably.
Do we need a regulator?
Question mark.
The medical industry, forexample, as a regulator to make
sure that people build medicalor medicines that don't harm
people.
Is it arguable that becausedata's become so prevalent and
(22:39):
so important in society, that weneed a regulator for a lot to
allow companies to behavehonorably and equitably
possibly, and probably.
Andreas Welsch (22:48):
Thanks.
So ethics obviously play plays akey role.
I know there is a lot of debatehappening between experts and
also on a government level, likeat the E.U.
with the AI Act or even overhere in, in the U.S.
But maybe question for you,building on that Somil what do
you see in your work?
(23:10):
How early should companiesreally think about ethics when
they do work around AI?
And who should actually own thistopic of AI ethics?
Somil Gupta (23:20):
Yeah, I focused a lot on the ethical
commercialization, ethicalmodernization.
The only kind of thing is that Ithink the ethics discussion
should not be an afterthought.
Ethics has to be built into thedesign of the product.
Before we even start talkingabout a product, we start
building the value framework.
And in that value framework, wehave to start embedding ethical
(23:41):
constraints and and ethicalpolicies.
The reason why this is importantis if you're not, put that in
the initial part of framing ofthe problem, both technical
framing and commercial framing,then what happens is that even
while you get your model up andrunning, it's totally too late
to look after ethics after themodel is being built.
(24:03):
And then we will be like tothink that our data scientists
are smart enough to figure outhow the model is responding, and
that is completely not true.
Once our model is deployed, ittends to have its own life and
how it's making prediction.
Of course, you could figure itout later on if you do a
diagnostics, but then it'salready too late.
(24:23):
Denial of service, infringing ofsomebody's rights manipulation
social engineering, all thoseare real threats now.
You cannot wait late enough.
You can't put early enough.
I can't stress upon this.
The framework that we create,the value framework.
In that value framework that wecreated the first time, that is
(24:43):
where we have to start lookingat ethics.
Because you start looking at whoare the people who are going to
be most vulnerable?
I think you give an example ofthe people who are elderly, who
might be impacted the most, butwho are the least who also the
most vulnerable.
We also saw some examples of gigeconomy, like drivers within
Uber who get impacted with analgorithm, on which they have.
(25:05):
Uber does not have any say onthis.
They can't figure out whetherthe decision was taken or was
right or wrong.
You have to understand that thedecision intelligence works
within the commercial framework.
You cannot control the decisionfrom point to point.
But we can definitely controlthe commercial framework which
we operate.
And we have to realize thatthese companies, they have
massive amounts of data,thousands of data points about
(25:25):
us.
They have a lot of power.
In what they can do.
So then really comes down tothis ethical design of
offerings, ethical pricing ofofferings.
Even though a risk like dynamicpricing.
So when we are doing dynamicpricing, we should not infringe
upon somebody's rights.
We should not discriminateagainst people.
So that's the kind of thingwhich is a trade off between
(25:48):
doing what is right versus doingwhat is easy.
And I hate to say that, but alot of companies, they're not
really inputting a thoughtaround these topics.
So all these ethics andeverything, governance, they
come very late in the cycle bythen, it's usually too late to
do anything about it.
Andreas Welsch (26:05):
Fantastic.
So when it comes to ethics,start early and make sure it's
pervasive throughout theproject.
That's awesome, team! We've madeit.
Thank you so much for helping meplay"What's the BUZZ?" today?
Kieran Gilmurray (26:18):
Hurray!
Andreas Welsch (26:20):
So let me quickly summarize.
First of all, if you think aboutdigital transformation, keep the
three Ps in mind (26:26):
people, place, and platform.
But really people should be thefocus and the center of your
initiatives, whether it's youremployees or your customers,
because the transformationcertainly impacts them in
different ways.
Secondly, if you're thinkingabout monetization of AI, really
(26:47):
don't treat it like any other ITproject, because it's not.
The different layers of it inand AI should really be core.
And as you're doing that, youget to number three, make sure
that ethics is a part of it fromthe very beginning, all the way
to the end of that project.
So we're getting close to theend of the show.
Somil Gupta (27:06):
Thanks, Andreas.
Popular Podcasts
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
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.