May 12, 2025 • 36 mins
In this episode of the Tech on Toast Podcast, Chris Fletcher is joined by JP Kloppers, Managing Director at Omnisient, to explore how hospitality operators can unlock the hidden value in their customer data—without compromising privacy.
JP explains how data collaboration works when it’s done responsibly, and how encrypted, anonymised data can create entirely new revenue streams, enable smarter AI adoption, and improve consistency across operations. Whether you’re running a multi-site group or managing a single site, this episode breaks down where AI and data intersect—and why this matters more than ever in a sector under pressure.
👉 Topics covered:
Why AI adoption in hospitality is still under 5%
What encrypted data collaboration actually looks like
The real-world value of restaurant CRM data
How to automate guest profiling and team support with AI
The role of large language models in streamlining operations
How retailers and banks are already collaborating—and why hospitality should follow
💡 “Every operator is sitting on a goldmine of data—they just don’t know how to activate it yet.” – JP Kloppers
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:08):
Hello and welcome to the Tech and Toast Podcast.
My name is Chris Fletcher and this is season 12.
Tech and Toast Podcast is serving up fresh chats with the
sharpest minds in hostility and tech.
If you're looking to level up your up, streamline your
service, or just sprinkle a little tech magic into your
business, you've come to the right place.
And guess what? Our partners Lightspeed are back
this season to look after us. They're the epos and payments
platform helping over 165,000 businesses.
(00:31):
Businesses such as Big Mama, Mildred Mallow, and Cubit House,
they'll help you skip scale that, you guessed it, the speed
of light. See what I did there?
With real time insights, faster transactions and slick
integrations with people like OpenTable, 0 News and many more,
Lightspeeds Unified Platform is built to boost your performance
and turn your tables quicker than you can say flat white
extra hot. So if you need fast funding,
(00:53):
check out Lightspeed Capital. No hidden fees, no interest
charges, just smart support whenyou need it.
If you're curious about all of this, just head over to
lightspeed.co.uk/tech on Toast and book yourself a free demo.
You won't regret it. Now enough of that, ready for
the next episode? Let's go.
(01:15):
Welcome to the Tech on Toast podcast, and today I'd like to
be joined by JP Kloppers, Managing Director at Omniscient.
That's right. So I get.
The reason I was saying it was some accent is because I was
trying to get it right. Is that did I get it right?
I'm listening, yeah. Perfect.
Welcome, JP, How are you? Very good, very good.
Thank you for having. Me.
You are welcome. And where are you based?
Are you based in London? I live just West of London, near
(01:35):
Reading, very far West of London, but we're based.
We're based in London. Yes, you're an office down here.
So you've travelled in just for the podcast or?
Got another meeting after. I don't know what it's like.
I've been inside. I haven't bothered going out
since lunchtime. It's beautiful out there.
Is it beautiful or too much? No, never too much.
You'll never hear me complainingabout the weather, the sunshine.
I'm out there and because you were telling me you're Dutch and
(01:57):
South African. South African.
South. African.
Dutch. And a long time back.
Dutch, yes. OK, well, welcome, Welcome to
the podcast. Before we get into talking about
omniscient and what you guys do there, let's just learn a bit
more about you. Tell me a bit about your
background, because you've got you've got a few marks on your
blots on your copy that I spotted.
I was like, I know those companies.
So yeah, tell us about your journey a little bit.
(02:19):
Yeah, that's right. So my background, I'm an
engineer, so a dangerous one at that.
Never worked as an engineer, always gravitated more towards
the the entrepreneurial journey.Came here seven years ago to the
UK with a business I was runningwhich at that stage were doing
social media analytics right, helping understand what's
driving why do customers love you?
(02:40):
Why do they hate you? Based on what was what what
they're saying? It's quite depressing, I
imagine. Well, no, actually it's.
Enlightening. Sorry.
It's, it's enlightening, it's, it's an unfiltered, raw version
of the truth and sometimes a little bit too true.
Anyway, we, we were helping brands then understand how to
position themselves accordingly.So that was what, what brought
(03:00):
me here originally. Then COVID hit and so we needed
to to pivot. So I then moved to Yoyo wallet,
which yes, which is where where I originally got into the
hospitality game. What was that like?
Cuz cuz you were you weren't in hostility prior to that.
It was kind of a more generic. Yeah.
So at that stage I, I joined as part of the acquisition Yoyo was
(03:21):
acquired by Taya. At that stage there called some
else but that doesn't matter. And they were rationalizing 2
loyalty payments and loyalty businesses.
There was yoyo here and there was a company in South Africa
called Y Group. Those two are both acquired at
the same time. And as part of bringing those
two businesses together, I came on board as the UKMD to to run
(03:43):
that merger, if you like both ofthe the businesses together and
then the businesses into into Taya.
Wow. And how was your first foray
into hospitality it? Was great met brilliant people I
think it's an industry which is just yeah, it's got great people
and just. I think it's we're quite
emotive, aren't we? We're quite, some would say hard
(04:03):
work, but but in a good way. Not hard work.
Yeah, I think everybody certainly at that, at that stage
and I think it's still at the moment is trying to do what's
best both for the business and for the customer.
And when you're at the technology side, there's just so
much room to do that well and todo it badly.
Yeah. It's just, I think that's a
struggle, isn't it, that those two choices that are doing,
(04:23):
doing what's best, because I think still the industry is kind
of working out what it wants to do.
Some people are have run down the tech Rd. quite far, some are
still very much at the start of that journey.
Thinking don't if I trust it yet.
That's right. And I mean the the other hat
that I wear is looking at AI adoption and how companies use
large language models and it's exactly the same.
You know, this is this powerful force that exists out there.
(04:46):
So is that something you do outside of your current role?
Or is that I've got 2 consultingjobs, I consult as the MD 4
amnesiant and then I also work as a as an AI adoption kind of
specialist. Looking at because what we did
in the social media world is we had a crowd of thousands of
people that worked on our platform to train our machines
and how to understand social media.
So my heart has always been in this world of how do people and
(05:10):
machines work well together? And date is obviously at the
intersection of those two. But what's now happened over the
past two years with the advent of large language models is it's
put that ability on steroids. And so suddenly companies are
thinking, what do we do, you know?
So on the omniscient side, we'vebuilt large language models into
the database so that a customer can easily query their data and
(05:32):
say, how many cans of soda have we sold in the last week?
And before you would have a datascientist that needs to answer
that question. Now you can just ask a large
language model that can go fetchthat data and tell you.
Wow, that's changed the because I didn't know that.
I mean, that's great. I've just found that out on the
podcast, which is really cool actually.
And I because I, I've got a bit of obsession with it as well.
(05:52):
No, I'm probably not as scientific about it as you are
with AI, but I am really on thatjourney of understanding where
all the helping the industry understand where they need to
get to or where that what it cando for them.
I think, I think that's, you know, like social media
actually, or like when they evenbefore that when the Internet
came, there was a period of timewhere people were like, I don't
need a website, why? I don't need a website or I
don't need social media. That's for kids, you know.
(06:14):
And then obviously it changes and people realise there's real
benefit to it. I think AI is that next kind of
change, yeah, where people are kind of on the cusp of it
thinking, OK, I, I now think therobots aren't coming yet, but I
understand that this thing can help me.
If you had to percentage eyes where you think I'm totally off
script, if you don't mind that. Is that all right?
Yeah, because someone else wrotethese probably anyway.
(06:35):
So if you had to percentage eyeswhere you think people are at
just generally with AI in terms of their 10% adopted and
understand it and totally can see the use cases, are we even
there yet? 5% yeah, yeah, I mean, I'm, I'm
doing some work with a listed company.
They've got 12,000 engineers wrestling with how do and for an
(06:57):
engineer, it's, it's a critical,critical thing to wrestle with
that how do you become more efficient and more effective at
your job? And that's at the very
forefront. And they, they even are trying
to figure this out. You know, you bring that back to
hospitality. It's this thing that's out
there. I like to think about it in the
kind of three stages. Everybody's looking at the
middle one, which is the large language model.
(07:18):
And every few weeks there's a new version of open IR releases
something and then you've got Sora and now you can do all
sorts of things. What people aren't thinking
about is how do we actually use our own data?
So for a for any hotel or restaurant or coffee shop,
they're sitting on a mountain oftheir own data, which they're
never going to put into a publicinto the public space, your
(07:41):
customer names, what they bought, but you want to be able
to train your own model on what that date is so that.
I was literally having this comment.
We had the operations director in this morning from a company
called The Other House, which isjust one property in London at
the moment. They're going to go to three
quite quickly. That's like a private members
residential club with rooms. And she was saying they use Muse
(08:02):
as their PMS and she was saying we've got to talk about AI
because every bloody episode does now, but we got to talk
about it. And she said actually she goes,
I'm not, she goes, I haven't gota clue about this stuff.
But I know now I'm used, I can ask about a customer.
So if JP was staying there, she can ask about JP and it'll tell
me everything about you, what you like to order, what
complaints you've had, what feedback you've given, what time
(08:24):
you'd like to arrive, What Car you'd like to arrive in, you
know, that kind of stuff. So she says I'm very like within
a touch of a button, we have an instant profile of a person
standing in front of us. She gives, which obviously years
ago we rely on the concierge or the full time receptionist who's
been there five years because now they're not there five years
anymore. 5 minutes is probably amore realistic life cycle for a
(08:44):
hospitality employee. So I think that's those use
cases I think are quite interesting that there's many,
many different ways it can be used.
But are you seeing people? Is it, is it?
Do they find that difference in applying it?
Is that is that what you think? We're so the difficulty is
getting your data into the model.
Yeah. So that's both in fine tuning
the model and then having some kind of a reference system that
the model can go back and find the information about the
(09:06):
customer. So that's come to the one end
and then on the other end would be the use case.
How does this manifest to the employee or the customer or the
manager or the end user? And that that's typical.
So to use an example, if you're a a lawyer, you're sitting with
years of cases that you've that you that you're not going to
give to somebody else because that's your IP.
(09:27):
Yeah. But to be able to quickly
reference that and say in this new case that I'm now looking
at, what are the 10 that are relevant to it and how does it
apply to this particular set of circumstances?
It's not just the the actual data, it's the the reasoning
behind the data as well potentially of matching that to
the subject. Yeah, that's a pretty good
(09:47):
exactly, yeah. Or, or if you're, I mean, I was
chatting to a wedding venue about two weeks ago and this,
they run a wedding venue out near where we live, 20
employees. They do around 22 events a
month. So a month.
A month. Yeah.
So a bit like fairly big scale. Yeah.
Yeah. But not big enough that they
have an IT department, Yes, but that maybe we'll just, you know,
(10:09):
as you do talk about what, what do you do?
What do I do instead of talking about this and talking about how
actually, if you have your own data that's in the large
language model, your employees now suddenly don't need to do
that. Mind numbing.
I've answered this question for the thousandth time.
Actually, if you plug it into your e-mail, it can look at this
question coming in. Here's a draft of how you have
(10:29):
responded every time before. You've now saved that employee
80% of their time, freeing them up to actually do the kinds of
things that is interesting to. Them it's crazy and the
potential impact I mean, I don'twant to get too lofty for the
world, but let's just stay in hostile potentially so that the
potential impact in terms of if you start with the real cynical
part, cost saving. So removing humans doing things
(10:52):
that they probably should be doing, which we've had to do
because there's no other way. Yeah, there's, there's AI
believe there's a huge, I mean, it's controversial because
obviously these people's jobs we're talking about and I'm sure
you can redeploy labour, but there is a big saving isn't
there to be had, particularly incentral, you know, like I
suppose I want to say fat, but areas where we're a bit, we're a
bit fat and we've got too much labour.
(11:12):
Yeah, Or where you've got, you know, a team of 10 people doing
one job. It's customer service.
Yeah, suddenly you can. You can do a lot of that work
using a machine. So either that those 10 people
now can go and do other things and you can have sort of higher
value work that they're deployedto.
Obviously it depends industry toindustry what that looks like.
(11:33):
But yes, on the cynical side, itcould be that people lose their
jobs. Yeah.
Well, it's the reality. I mean, I think it's the reality
of of of new things that come, that things change, right.
And that's right. The smart people see that change
and coming and kind of yeah, move to adapt to it.
And I think for businesses that are seeing that opportunity
because it is every crisis or every new thing is an
opportunity. And my deep conviction is that
(11:55):
people matter, and machines should make us better, not
worse. Yeah, So.
And that, that's the future I want to fight for.
Yeah. And I think in our industry
particularly that everyone's in that on that boat.
I think most people are. I think the one elephant in the
room is that the headwinds are constantly bad in terms of
economy. It's constantly a challenge.
And I think if there's an opportunity to save money
(12:18):
because we can do something better.
And I think the big thing here is the one thing that no one
really talks that AI is consistency, that reality that
you're going to get the same information from me and from,
well, from me, the robot, every time I give it to you, no matter
what, no matter what day it isn't.
If I've had a bad day, whatever it is, you're going to have the
same information. I think that's not the same with
humans. And I know in certain senses we
(12:39):
like that. But when you're providing
compliance or health and safety or whatever it might be, you
want to be consistent, flat information.
Yeah. Yeah, that's a good point.
Yeah. You know, telling pilots what to
do with a storm coming. Yeah.
But but it's true, isn't it, that I don't think that I think
they're consistent. When I was in operations, I
spent every managers exist, I think because we haven't got AI
(13:03):
or because, I mean, they're probably listening and thanks
mate. But when I was in every manager,
my job is really just to manage the information around me and
make sure the team's got it. Obviously there's a bit more
around that I was looking after individuals, but.
But I, I think that middle management is going to be
radically impacted by it becauseyou can now have, you know, in
the system that I've described where you've got your own data,
(13:23):
you can put in your own performance information.
These are key objectives for thenext year that's broken down by
quarter. This is by team, this is by
individual. Then you can have a system
that's listening to every meeting, which we have, you
know, the doctors of the world that can automatically then go,
OK, how are you? How is this team tracking
against their key objectives? The manager no longer needs to
(13:44):
do endless performance and they hate it.
They hate it, yeah. But that the machine can put all
of that together and say here's how we're tracking and then you
can have the conversation that you need to around, OK, how do
we course correct? These things have have now
emerged. We're struggling in this area.
Like what do we need to do? So those kinds of things, I
think are just going to make people way better if they use
the tools. Well, if they don't then.
(14:05):
Well, I think it's the same as everything, right?
Garbage in, garbage out. And I think you mentioned data
before. The risk is for all these people
that have data, how good is thator relevant is that data or how
clean is it? And when I say clean, I mean,
you know, like did that? Has that customer left them?
You know, are they going to get put into the mix?
Because my worry is that they'regoing to Chuck all of that data
into some bot and they're going to do something to stop plowing
(14:27):
out stuff to people that don't even love them anymore or don't
want to hear from them. So there's a bit of that going
on, but I think that's the same as any use of data, right?
You need to clean up. I need to talk about omniscient.
Because I this is a good take away.
Yeah, because I, because I deliberately got excited about
AI because I am really invested in it and wants to know what's
going on. Tell me a little bit about what
it does and how it represents. And obviously it's quite closely
(14:48):
linked to what we just spoke about a data piece anyway.
That's right, yeah. And and at the core of it is.
A That's why I did it, JP. I was deliberately, I
deliberately LED, LED you down apath.
Thank you. Thank you.
Not well. Done.
So what Omniscient does, It's a data collaboration platform that
allows two or more parties to share their data.
And I use share in inverted commas because core to the
(15:10):
platform is that nobody ever gives their data to somebody
else. Yes.
But the kind of core use case that we were born out of was
allowing a bank to build a credit model from grocery data.
So where you know, I'm just moved here.
The Experian or TransUnion know nothing about me.
The bank won't give me credit. Not even a telco will give me a
(15:32):
cell phone because they don't they don't know who.
I am. You've got no background.
And it takes a long time to build up that credit profile,
few years. I think it was about three years
before I could get a cell phone contract.
JP was quiet. Yeah, exactly.
Where's your grocery data? That's very, very and recent and
I suppose highly predictable. Yeah, I was going to say
(15:54):
financial capacity. A lot of patterns emerge quite
quickly. Yeah.
And what we've learned, so the kind of use case I understand
was born in South Africa and expanded here, headquartered in
the UK now. And what we learned in South
Africa is that that grocery datais highly predictive of
financial capacity. And so there where they were,
(16:14):
the big retailer that we work with is called ShopRite.
They had 8 million customers that the banks didn't know who
they were. No, no credit information about
them using their grocery data, they were able to see, hang on,
three million of these should bein the financial system.
Wow, they can afford credit. It's not going to be bad for
them. They can kind of get into their
(16:35):
foot into the financial, you know, whether it's buying a
house or getting a loan for a car, like all of these things
help people progress in their lives.
But if you have no credit score,you don't have access to that
system. It's crazy.
And so this, this data collaboration enabled the bank
to appropriately and responsiblylend to customers that could
afford it. For the customer, they suddenly
(16:56):
like, wow, I can now when I've been denied before I've applied
for a loan to buy a house, the bank said no, because they don't
know who I am. Now I can use my grocery data to
do that. So that that's kind of the the
hero use case, if you like, that, I understand was born out
of as that world has has evolved.
So the platform has become more and more useful to allow maybe
(17:18):
not non endemic brands to collaborate.
So in the hospitality space, youwould have a car rental company
collaborating with a, a hotel. You know, that's, that's an
obvious connection. What you wouldn't have is an
insurer collaborating with a hotel to say or with the the
airline to say let's provide travel insurance.
(17:38):
And I saw I don't know this total side but but on Dragon's
Den there's ATVI don't know if you know the TV show Dragon's
Den, there was a young 2 girls on there who'd built something
for rental. Same story people couldn't rent
properties in London because they had no background and they
had no credit history and they were building exactly same
thing. So they were building trying to
(17:59):
build a credit history based on their other spending to allow
them. So a very similar story, I think
to allow them access to that. It got Pooh poohed.
There was something in it that they were saying that landlords
wouldn't buy into it or whatever.
It was a company it was but but it was a very similar story and
and just out of interest on on the South African side of things
did is that still is that still going?
Is that something that. Yeah.
So they they. Use that.
(18:20):
So in South Africa now most of the retailers on the platform,
all of the banks, all of the insurers, medical aid companies,
health companies as the and thisis where the kind of beautiful,
the beauty of this ecosystem comes into sort of a league of
its own. We have a concept of a data
seller. So that would be the retailer in
our conversation, it might be the restaurant or the hotel,
(18:43):
somebody who has customer data and says we think this is
valuable to somebody, make it work.
What the platform allows them todo is to put it into the
platform in a completely anonymous way, so they encrypt
any customer data before it leaves their environment.
So even if somebody hacked Omniscient, they would never get
access to their customer data, never left.
That's why I use share and enriched commas because it's
(19:05):
encrypted. So once it's in our platform,
even the grocer doesn't know who's who that customer is.
But the bank or the insurer can look at that data and put their
own dates in and do a distinct match at a customer level.
But nobody knows who the customer is because it's all
encrypted. So the bank and then look at
that. So let's say a restaurant had
their dates in there and the bank is saying we want, we've
(19:28):
got a, a credit card offer for high net worth individuals.
As an example, they could look at the customer data from the
restaurant loyalty program and DDupit say, well, actually these
10,000 customers are already ours.
But actually there are another. If we look at these 10,000
customers that we've got, these 2000 are good customers to us.
(19:49):
Who else behaves like these two,these 2000 customers at the at
the restaurant and they can thengo to the restaurant with an
offer to say, we'd like to partner with you.
We're prepared to pay you X to do a combined offer to your
customers. And once they, once all parties
agree, then the restaurant can then take that offer to their
customers. But they're always in control to
(20:10):
say, is this in the best interest of our customers?
Do we have the right to use their data in this way?
You know all of those things they.
But the anomalous part makes it so unique, doesn't it?
Because the fact that you can goalong and say, look, here is my
bunch of data. This is what it, this is what
we've got. And because the fact is they're
sitting on the state that has value that they probably don't
know about. That's right.
Or they're not thinking about the value that it has maybe.
(20:32):
By the way, it's not their business.
Yeah, and it's similar for the retailer.
The, you know, a grocery retailer's not thinking about
why is this our data valuable toa bank or an insurer and what's
the Amnesium platform allows them to do is to at no human
cost to them, at no risk to their customers and at no
expense put their data into a platform.
(20:52):
It's a no brainer. I think you might have the best
job ever. The.
Bank and then their data scientists look at this data and
say and we and that's the role of Nissan plays is to bring both
the buyer and the seller together.
Almost like the broker. Yeah, Data broker type.
Thing and in that environment have the we call it a you know a
machine learning environment Jupiter notebooks as the
(21:12):
platform where a data scientist can then build quite a
complicated look alike model if they want to which which takes
the credit information from the bank or the insurer overlays
that with the restaurant or the retailer or the grocer or
whoever they buy the seller is and then look at OK what kind of
offers can we come up with out of this data they can never
(21:33):
action it all they can do is go to the data seller and say we'd
like to do this because you've got yeah once the once they
agree then it's then they can you know and.
I probably should bring this in.What?
What do the commercials look like?
I mean, what potentially, you know, what kind of offer would
you think of with a card? Provider, It depends very much
on what that offer is, so if it's a mortgage matter, they'd
(21:58):
probably be prepared to pay £1000 to bring you a mortgage.
Customer, If it's a credit card,maybe a hundred 200 lbs who
knows. I told is this like data mining?
But someone was talking about they do, they've got a platform
where they're data mining and they were offering to
restaurants and maybe, I don't know if I understood him
properly, but they're offering to restaurants.
So sort of the similar thing saying look, you've got a bunch
(22:20):
of data you're sitting on, give it to us, we'll mine it,
whatever that means. And then we'll unearth diamonds
as it would be, or unearth quality for.
You value for you. You could use that term.
I think the the key thing in theAmnestian platform is they never
give up control, right? So the danger with giving your
data to somebody that's going tomine it is you relinquish your
(22:41):
IP. Yeah.
So someone, I mean number regardless of what they say,
someone could quite easily. Yeah, it could be or they find
value and then they can action it.
Yeah. Where once you've encrypted it
so you know the different degrees of you sharing or
collaborating or mining that data.
A key thing is that encryption layer where you've never given
up your IP, you've never given up access and you've never given
(23:03):
up control. And at any stage they can remove
any of those to say we we no longer comfortable with this.
Partnership and and is there is there a common theme when
they're doing when not just necessary hostility, but is
there a common theme when peopleare owning their IP and giving
up the data giving up something the right word but sharing yeah
I've got my an invasive camera yeah sharing their data.
(23:25):
Is there any kind of commonalitythat that that comes out of that
when they discover what they've shared, that they they're like,
oh, wow, that wow moment. They're like, OK, we've actually
got some really useful data herethat could absolutely.
Yeah, yeah. Is it very rare that you don't
get that? If if you don't have the
counterparty, So when the kind of ecosystems in its early
stages have you, you need buyersand sellers, yes.
(23:46):
And if you only have one buyer and they're not interested,
then. That's very true.
That's a crap deal. Yeah, exactly.
But but it the the the other like lens to look at it through
is for a restaurant or a hotel, they would have their own bank
that they bank with. So that that would be another
area where they can collaborate where currently you can't, you
(24:09):
can't upload your customer data to the bank and say who else at
the bank looks like our great customers.
Let's do a partnership with our own retail bank.
Just say to put an offer to other retail bank customers to
say come to our restaurant, cometo our hotel, Let's let's do
some. Is there like a bit, is this
quite a lot of an education piece then for the people's data
(24:30):
you want to share because it's because.
That's a bit a big part of a bigpart of the job is just talking
about the art of the possible and how everybody knows that
data is valuable, everybody's nervous about giving it away.
And so bringing those two together.
So I think education is key. You know, it's why we engage the
regulator. For the regulator, this is
critical to have the privacy of the individual protected while
(24:53):
not stifling the economy. The number one is their flag in
the sand, isn't it? Yeah.
And that's what omniscient does.It brings those two things
together to say we're not going to compromise somebody's
privacy. That's that's sacrosanct.
But we also want to create a data economy where businesses
can thrive. Yeah, it's having a responsible
platform to do that. And.
It does seem crazy, doesn't it, that we sit on all this data and
(25:13):
then it doesn't people see value.
I mean, sometimes in hostility, most people talk about CRM data
as e-mail marketing or sending something out on your birthday
or whatever it might be to get areturn.
But actually this is a very different conversation.
This is about the demographic ofyour business or the people
there's because there's different types of data.
Am I psychographic? Am I right in saying there's
different types of data that youcan get out of that, all
(25:35):
different types of learning you can get out of it and I think
understanding those are really key for everything, right, for
marketing purposes or what? You might want to do exactly.
I mean, if you, if you're running a restaurant, how
valuable would it be to go to your bank and know that actually
you've got customers who might live close to you or live far
away, That doesn't matter. But it would be great to know
(25:57):
that. Do your customers also go to
other restaurants? Yeah, What types of restaurants
do they go to? Where else do they spend their
money in an anonymous way? These are this is easy to share
and collaborate. Absolute gold.
Yeah, because. The bank's got that information.
Yeah. There's no mechanism to to
easily kind of reach. Well, I haven't.
Heard of you before. Reach that insight if you like
(26:18):
both parties. So while initially there's kind
of a clear seller and buyer in the relationship, quite often
those roles get reversed. We're actually now the
retailer's interested in the bank's data just as much as the
bank's interested in the in the retail.
And you're right, you'd never, why would you ever partner with
your bank? I mean, I just can't think of
any apart from to get a loan or and that's not really a
partnership, is it? It's kind of a one way
(26:40):
partnership that but yeah, but there's no other reason why you
would. And I think that again, it'll
build some structure around thatrelationship as well as, you
know, find out some true value in in what you've got that
you're already sitting on and what kind of numbers you look,
because I think they're sitting on a lot of data restaurants.
I think, I mean on average, I mean someone was talking to me
the other day and we're talking hundreds of thousands of emails
(27:01):
and whatever or customers that they've had through the doors
or. Yeah, yeah, yeah, two different.
I mean the obviously the coffee chains like the Neuros,
Starbucks, those are in the millions of customers, a
restaurant chain with 100 sites,they're probably sitting on a
few 100,000 customers that have depending if they have a loyalty
program or not. Obviously if they don't, they
(27:21):
won't know who those customers are.
But then again, if you've got card data you could part with
your bank, there are ways in which they can look at how to
kick start that sort of data economy that they can benefit
from and then you're appropriately.
Partnering and your AI knowledgeor your expertise in AI you'll
then because that's actually built into omnisci and obviously
(27:42):
the in the middle that matched the matching part is done by AI
presume or the yeah. We've got a smart machine
learning system that will look at if one customer's got e-mail
address, first name, post code, somebody else has got cell phone
post code and some other piece of data.
There are different ways that you can match that data to try
(28:02):
and look at, you know, what is the intersection of these
customers, but that's, you know,we'll be getting into a lot of
technology down that path and. Then, and you go through that
posters then and obviously there's some value to the data
that the customer is now aware of.
And the next step is that they deal with the partner, they deal
with the bank, whoever it might be, they do some kind of deal
(28:23):
and off they go. Yeah.
The the process is customer data.
Seller puts the data into the platform.
Yeah, we would then have our data scientists look at that
data and see what where do we think the value here is?
And what's that? Is that a long?
I mean, I hate to put a timeline.
Are they looking at a long period of time?
How long does that take them from meeting you and giving you
my sharing my data? I've got to find.
(28:44):
You've got to find a new word for sharing.
A new word? Yeah, Lending depending on
depending on who it is, if it's if it's a tier one bank or
retailer, they take a long time getting the data in a simple
getting over privacy and security is.
So much a bit more difficult. Yeah, yeah.
Because they want to know is it safe?
All all the right questions. But we've, you know, we've done
(29:05):
this globally with TransUnion, which is probably the hardest
way anybody can get through to have a Bureau that's prepared
to, to collaborate with their data.
So we know we can answer all those questions.
It's just a question of, you know, going or going through the
jumping through the hoops. Once you through that then it's
quite quick, you know getting data in as a few weeks.
And how is it delivered back to the customer and what, what kind
(29:27):
of how is it presented back to the customer in terms of what
you've found out? Do you know what I mean?
So as in I'm the MD of Chris's restaurant chain.
You've looked at my data. Do I get back some kind of
report that says? There would be a report, but
more often than not we would come with a proposition to say
we think this ensures. Right.
OK. So you would start that matching
(29:48):
process, right? OK.
That that's so omniscient. There are three things that make
make us unique when you put themtogether. 1 is that privacy
piece. Yeah, the, the, the
cryptographic encryption. 2 is the matchmaking bit that you're
talking about now. And #3 is that machine learning
platform that allows the buyer to create a quite a
(30:09):
sophisticated model out of the seller's data and look at the
intersection of them. So that works.
That was talking about. That is pretty much not done,
but by the time you've gone through the process of looking
at the data, you've done the matching or proposed the
matching. Yeah.
And and quite often we'll have another customer that's got a
similar data set that we've worked with before.
Yeah, I imagine that's quite common.
Yeah, imagine. You know, OK, this kind of
(30:30):
frequency of transaction with these sorts of purchases.
That could work with. Could be interesting to.
Travel insurance or pet insurance or credit card or
debit or you know what, whateverthe and.
What was it? Was it wasn't born out of
hostility that it was born out of?
Born out of the credit space, right?
That's right. The founders come out of the
more, more the banking side. So I think that that was the
missing pieces. Retailers for a long time have
(30:52):
been selling their data to theirsuppliers.
Yes. You know, the unilevers of the
world. Yeah, exactly.
Yeah. What they weren't doing is, is
selling that data to the banks. Yeah, but when you come from the
bank world, the banks for many, many years been trying to get
their hands on grocery data because they know it's so.
It's so rich. Yeah, it's resilient, I suppose,
(31:14):
because it is the most consistent thing a human will
do, right? Exactly.
It's very current, yeah. You know, right now the case
somebody started buying eggs. Therefore, the, you know, we've
all heard the stories about the,you know, when the retailer knew
before the dad did that the daughter was pregnant because of
what she was buying. Yeah.
And we've all heard those stories.
But that that's true. You know those those things are.
(31:36):
That's the, I think that's the really fascinating part about
it, how you can look at moments in people's lives potentially
without knowing the person, without knowing the moment,
yeah, if you know what I mean, without actually knowing what
happened and see changes, changes in spending, Yeah.
And changes in behaviour, which I think is so crucial for.
That's right. So the, the retailers have known
this, you know, Tesco's been doing this for 20 years.
(31:59):
There's no ways they would ever give that data to to a bank or
an. Insurance.
That's a chance, no? They'd be compromised by the
customers, you know, the integrity they have in the
relationship that they've built for 20 years, the.
Most successful loyalty program on the planet.
And so having a way in which they can let a bank look at it
without ever giving anything away and the bank can come and
say this is what we want to do for your customer.
(32:19):
And the, and the retailer says, is this in the best interest of
our customer? If they think, yes, it is, but
hey, let's let's collaborate. If they think no, they say,
sorry, we're not, we're not interested in in that, whatever
the use case is. And we've had many of those
examples where you know, the bank finds value or the insurer
does and. The side says you know if.
(32:40):
You know, if somebody's going todefault on their credit card
long before they default, if they're spending patterns change
or if they start buying lots of alcohol, you know, down the
line, there's going to be an insurance claim.
Yeah, But that's not in the bestinterest of the customer, No.
So the retailer would say, hang on, like while that we know that
data can be used for that, we don't want it to be used for
that. It's.
Very cool. And and and this is all insulin
(33:03):
I suppose with when we look at AI and everything we talked
about at the start of the podcast.
And I am interested to see whereyou think that might go in terms
of because as you mentioned, there's lots of GBTS and chat
GBTS or whatever, different versions coming out everywhere.
Where, where do you think it will, where do you think we'll
end up and how quickly will we end up there in terms of AI and
(33:23):
what and we talked about a little bit at the start, not
just in hospitality, but in general day-to-day use.
Do you think because I feel likeeverybody's kind of on ChatGPT,
maybe I'm not right. Actually my, my wife isn't, but
you know, but I feel like a lot of my work friends are all using
it now and kind of use it regularly.
Where do you think it might end up?
I mean, on the AI side, I think the, I think we're going to go
(33:44):
through a few waves. I think the next one is how do
teams work together? We've got individuals working,
but it hasn't yet gotten to a stage where companies have a
united kind of way of dealing with.
Like an interface where we're all.
Yeah, together. Yeah.
So that that would be, I think the next one, like a team of
engineers is a good example you have and individuals figuring
out how this makes them better. You don't yet.
(34:07):
We're not yet at the stage whereteams are looking at how it
makes that whole team more effective, so I think that
that's probably the next. Yeah, that's where I actually
had an instant the other day when we were talking about
something between our team and we were trying to take some
data, weirdly, maybe I should give you a ring.
We were taking about 60,000 pieces of data and trying to, we
were trying to work it through ChatGPT and chunks to see if it
(34:27):
could do much with it. And it was quite interesting
actually. It got, but then we were like,
we weren't using the front end of the data we were using there,
the systems they use the tech systems.
So we're trying to get some pattern out of it, but it kept
breaking. Yeah.
So so this is a classic example where having your own private
large language model that is trained on your data can
reference your data. Exactly that.
(34:48):
That, and to me, that's the nextstep.
Every company is going to get there from the smallest to the
largest. Yeah.
Having a private space where their own, you know, their own
machines, whether that's in the cloud or on their own hardware
doesn't matter. Needs to hurry up.
I mean, I, I genuinely believe that's one of the single such an
unlocking moment for a lot of companies because I think a lot
of companies just especially in hostility, we just kind of
(35:10):
bounce through day-to-day and wekind of just get there, get
there. And then to have that power of
understanding why we're doing what we're doing, what results
we're having, and kind of where the successes and failures are
or where the opportunities are without having to do too much
Excel work would be absolutely groundbreaking.
You know, and and there, once you've got those together, then
(35:31):
the number of use cases you can build on top of that is endless.
Yeah, you know from a simple, just your own private ChatGPT
interface that you can query. And then?
Tell me about this to a chat botthat sits on your website that
can answer customer specific questions or company specific
questions. To the performance management
one we spoke about that middle managers can use to track how
(35:54):
people are doing and have the right conversations.
It's exciting. I mean, I think it's brilliant.
I mean, I think I've only reallygot involved.
I worked with a guy called Patrick who's a large language
model expert, like a grad from Oxford, super smart chap, a lot
like yourself. And he and he, he talks a lot
about, you know, where it's going and what he's doing.
And he's built some incredible stuff with it already.
(36:15):
And I, yeah, I think it's an opportunity that everyone will
be is already grasping, but we'll probably be there quite
quickly. Look, we only get half an hour
with you people and you're superinteresting.
What's the best way to get hold of you?
If I can, if I can send in my listeners your way.
Do you have time for two? On the omniscient side, it's
omniscient.com, right? And for the AI consulting, it's
adaga.com. Adaga, very nice.
(36:36):
Is that your, is that your business?
That's my business. That's your consulting business.
Very cool. Look, Thank you very much for
coming in. Enjoy the sun.
We've actually got 24°. That is something.
It's beautiful. Yeah, well, I wouldn't know.
I'm in a dark room with their cups on the wall.
So that was that was JP everybody.
And we shall see you next week. Thanks, JP.
Hello and welcome to the Tech and Toast Podcast.
My name is Chris Fletcher and this is season 12.
Tech and Toast Podcast is serving up fresh chats with the
sharpest minds in hostility and tech.
If you're looking to level up your up, streamline your
service, or just sprinkle a little tech magic into your
business, you've come to the right place.
And guess what? Our partners Lightspeed are back
this season to look after us. They're the epos and payments
platform helping over 165,000 businesses.
(00:31):
Businesses such as Big Mama, Mildred Mallow, and Cubit House,
they'll help you skip scale that, you guessed it, the speed
of light. See what I did there?
With real time insights, faster transactions and slick
integrations with people like OpenTable, 0 News and many more,
Lightspeeds Unified Platform is built to boost your performance
and turn your tables quicker than you can say flat white
extra hot. So if you need fast funding,
(00:53):
check out Lightspeed Capital. No hidden fees, no interest
charges, just smart support whenyou need it.
If you're curious about all of this, just head over to
lightspeed.co.uk/tech on Toast and book yourself a free demo.
You won't regret it. Now enough of that, ready for
the next episode? Let's go.
(01:15):
Welcome to the Tech on Toast podcast, and today I'd like to
be joined by JP Kloppers, Managing Director at Omniscient.
That's right. So I get.
The reason I was saying it was some accent is because I was
trying to get it right. Is that did I get it right?
I'm listening, yeah. Perfect.
Welcome, JP, How are you? Very good, very good.
Thank you for having. Me.
You are welcome. And where are you based?
Are you based in London? I live just West of London, near
(01:35):
Reading, very far West of London, but we're based.
We're based in London. Yes, you're an office down here.
So you've travelled in just for the podcast or?
Got another meeting after. I don't know what it's like.
I've been inside. I haven't bothered going out
since lunchtime. It's beautiful out there.
Is it beautiful or too much? No, never too much.
You'll never hear me complainingabout the weather, the sunshine.
I'm out there and because you were telling me you're Dutch and
(01:57):
South African. South African.
South. African.
Dutch. And a long time back.
Dutch, yes. OK, well, welcome, Welcome to
the podcast. Before we get into talking about
omniscient and what you guys do there, let's just learn a bit
more about you. Tell me a bit about your
background, because you've got you've got a few marks on your
blots on your copy that I spotted.
I was like, I know those companies.
So yeah, tell us about your journey a little bit.
(02:19):
Yeah, that's right. So my background, I'm an
engineer, so a dangerous one at that.
Never worked as an engineer, always gravitated more towards
the the entrepreneurial journey.Came here seven years ago to the
UK with a business I was runningwhich at that stage were doing
social media analytics right, helping understand what's
driving why do customers love you?
(02:40):
Why do they hate you? Based on what was what what
they're saying? It's quite depressing, I
imagine. Well, no, actually it's.
Enlightening. Sorry.
It's, it's enlightening, it's, it's an unfiltered, raw version
of the truth and sometimes a little bit too true.
Anyway, we, we were helping brands then understand how to
position themselves accordingly.So that was what, what brought
(03:00):
me here originally. Then COVID hit and so we needed
to to pivot. So I then moved to Yoyo wallet,
which yes, which is where where I originally got into the
hospitality game. What was that like?
Cuz cuz you were you weren't in hostility prior to that.
It was kind of a more generic. Yeah.
So at that stage I, I joined as part of the acquisition Yoyo was
(03:21):
acquired by Taya. At that stage there called some
else but that doesn't matter. And they were rationalizing 2
loyalty payments and loyalty businesses.
There was yoyo here and there was a company in South Africa
called Y Group. Those two are both acquired at
the same time. And as part of bringing those
two businesses together, I came on board as the UKMD to to run
(03:43):
that merger, if you like both ofthe the businesses together and
then the businesses into into Taya.
Wow. And how was your first foray
into hospitality it? Was great met brilliant people I
think it's an industry which is just yeah, it's got great people
and just. I think it's we're quite
emotive, aren't we? We're quite, some would say hard
(04:03):
work, but but in a good way. Not hard work.
Yeah, I think everybody certainly at that, at that stage
and I think it's still at the moment is trying to do what's
best both for the business and for the customer.
And when you're at the technology side, there's just so
much room to do that well and todo it badly.
Yeah. It's just, I think that's a
struggle, isn't it, that those two choices that are doing,
(04:23):
doing what's best, because I think still the industry is kind
of working out what it wants to do.
Some people are have run down the tech Rd. quite far, some are
still very much at the start of that journey.
Thinking don't if I trust it yet.
That's right. And I mean the the other hat
that I wear is looking at AI adoption and how companies use
large language models and it's exactly the same.
You know, this is this powerful force that exists out there.
(04:46):
So is that something you do outside of your current role?
Or is that I've got 2 consultingjobs, I consult as the MD 4
amnesiant and then I also work as a as an AI adoption kind of
specialist. Looking at because what we did
in the social media world is we had a crowd of thousands of
people that worked on our platform to train our machines
and how to understand social media.
So my heart has always been in this world of how do people and
(05:10):
machines work well together? And date is obviously at the
intersection of those two. But what's now happened over the
past two years with the advent of large language models is it's
put that ability on steroids. And so suddenly companies are
thinking, what do we do, you know?
So on the omniscient side, we'vebuilt large language models into
the database so that a customer can easily query their data and
(05:32):
say, how many cans of soda have we sold in the last week?
And before you would have a datascientist that needs to answer
that question. Now you can just ask a large
language model that can go fetchthat data and tell you.
Wow, that's changed the because I didn't know that.
I mean, that's great. I've just found that out on the
podcast, which is really cool actually.
And I because I, I've got a bit of obsession with it as well.
(05:52):
No, I'm probably not as scientific about it as you are
with AI, but I am really on thatjourney of understanding where
all the helping the industry understand where they need to
get to or where that what it cando for them.
I think, I think that's, you know, like social media
actually, or like when they evenbefore that when the Internet
came, there was a period of timewhere people were like, I don't
need a website, why? I don't need a website or I
don't need social media. That's for kids, you know.
(06:14):
And then obviously it changes and people realise there's real
benefit to it. I think AI is that next kind of
change, yeah, where people are kind of on the cusp of it
thinking, OK, I, I now think therobots aren't coming yet, but I
understand that this thing can help me.
If you had to percentage eyes where you think I'm totally off
script, if you don't mind that. Is that all right?
Yeah, because someone else wrotethese probably anyway.
(06:35):
So if you had to percentage eyeswhere you think people are at
just generally with AI in terms of their 10% adopted and
understand it and totally can see the use cases, are we even
there yet? 5% yeah, yeah, I mean, I'm, I'm
doing some work with a listed company.
They've got 12,000 engineers wrestling with how do and for an
(06:57):
engineer, it's, it's a critical,critical thing to wrestle with
that how do you become more efficient and more effective at
your job? And that's at the very
forefront. And they, they even are trying
to figure this out. You know, you bring that back to
hospitality. It's this thing that's out
there. I like to think about it in the
kind of three stages. Everybody's looking at the
middle one, which is the large language model.
(07:18):
And every few weeks there's a new version of open IR releases
something and then you've got Sora and now you can do all
sorts of things. What people aren't thinking
about is how do we actually use our own data?
So for a for any hotel or restaurant or coffee shop,
they're sitting on a mountain oftheir own data, which they're
never going to put into a publicinto the public space, your
(07:41):
customer names, what they bought, but you want to be able
to train your own model on what that date is so that.
I was literally having this comment.
We had the operations director in this morning from a company
called The Other House, which isjust one property in London at
the moment. They're going to go to three
quite quickly. That's like a private members
residential club with rooms. And she was saying they use Muse
(08:02):
as their PMS and she was saying we've got to talk about AI
because every bloody episode does now, but we got to talk
about it. And she said actually she goes,
I'm not, she goes, I haven't gota clue about this stuff.
But I know now I'm used, I can ask about a customer.
So if JP was staying there, she can ask about JP and it'll tell
me everything about you, what you like to order, what
complaints you've had, what feedback you've given, what time
(08:24):
you'd like to arrive, What Car you'd like to arrive in, you
know, that kind of stuff. So she says I'm very like within
a touch of a button, we have an instant profile of a person
standing in front of us. She gives, which obviously years
ago we rely on the concierge or the full time receptionist who's
been there five years because now they're not there five years
anymore. 5 minutes is probably amore realistic life cycle for a
(08:44):
hospitality employee. So I think that's those use
cases I think are quite interesting that there's many,
many different ways it can be used.
But are you seeing people? Is it, is it?
Do they find that difference in applying it?
Is that is that what you think? We're so the difficulty is
getting your data into the model.
Yeah. So that's both in fine tuning
the model and then having some kind of a reference system that
the model can go back and find the information about the
(09:06):
customer. So that's come to the one end
and then on the other end would be the use case.
How does this manifest to the employee or the customer or the
manager or the end user? And that that's typical.
So to use an example, if you're a a lawyer, you're sitting with
years of cases that you've that you that you're not going to
give to somebody else because that's your IP.
(09:27):
Yeah. But to be able to quickly
reference that and say in this new case that I'm now looking
at, what are the 10 that are relevant to it and how does it
apply to this particular set of circumstances?
It's not just the the actual data, it's the the reasoning
behind the data as well potentially of matching that to
the subject. Yeah, that's a pretty good
(09:47):
exactly, yeah. Or, or if you're, I mean, I was
chatting to a wedding venue about two weeks ago and this,
they run a wedding venue out near where we live, 20
employees. They do around 22 events a
month. So a month.
A month. Yeah.
So a bit like fairly big scale. Yeah.
Yeah. But not big enough that they
have an IT department, Yes, but that maybe we'll just, you know,
(10:09):
as you do talk about what, what do you do?
What do I do instead of talking about this and talking about how
actually, if you have your own data that's in the large
language model, your employees now suddenly don't need to do
that. Mind numbing.
I've answered this question for the thousandth time.
Actually, if you plug it into your e-mail, it can look at this
question coming in. Here's a draft of how you have
(10:29):
responded every time before. You've now saved that employee
80% of their time, freeing them up to actually do the kinds of
things that is interesting to. Them it's crazy and the
potential impact I mean, I don'twant to get too lofty for the
world, but let's just stay in hostile potentially so that the
potential impact in terms of if you start with the real cynical
part, cost saving. So removing humans doing things
(10:52):
that they probably should be doing, which we've had to do
because there's no other way. Yeah, there's, there's AI
believe there's a huge, I mean, it's controversial because
obviously these people's jobs we're talking about and I'm sure
you can redeploy labour, but there is a big saving isn't
there to be had, particularly incentral, you know, like I
suppose I want to say fat, but areas where we're a bit, we're a
bit fat and we've got too much labour.
(11:12):
Yeah, Or where you've got, you know, a team of 10 people doing
one job. It's customer service.
Yeah, suddenly you can. You can do a lot of that work
using a machine. So either that those 10 people
now can go and do other things and you can have sort of higher
value work that they're deployedto.
Obviously it depends industry toindustry what that looks like.
(11:33):
But yes, on the cynical side, itcould be that people lose their
jobs. Yeah.
Well, it's the reality. I mean, I think it's the reality
of of of new things that come, that things change, right.
And that's right. The smart people see that change
and coming and kind of yeah, move to adapt to it.
And I think for businesses that are seeing that opportunity
because it is every crisis or every new thing is an
opportunity. And my deep conviction is that
(11:55):
people matter, and machines should make us better, not
worse. Yeah, So.
And that, that's the future I want to fight for.
Yeah. And I think in our industry
particularly that everyone's in that on that boat.
I think most people are. I think the one elephant in the
room is that the headwinds are constantly bad in terms of
economy. It's constantly a challenge.
And I think if there's an opportunity to save money
(12:18):
because we can do something better.
And I think the big thing here is the one thing that no one
really talks that AI is consistency, that reality that
you're going to get the same information from me and from,
well, from me, the robot, every time I give it to you, no matter
what, no matter what day it isn't.
If I've had a bad day, whatever it is, you're going to have the
same information. I think that's not the same with
humans. And I know in certain senses we
(12:39):
like that. But when you're providing
compliance or health and safety or whatever it might be, you
want to be consistent, flat information.
Yeah. Yeah, that's a good point.
Yeah. You know, telling pilots what to
do with a storm coming. Yeah.
But but it's true, isn't it, that I don't think that I think
they're consistent. When I was in operations, I
spent every managers exist, I think because we haven't got AI
(13:03):
or because, I mean, they're probably listening and thanks
mate. But when I was in every manager,
my job is really just to manage the information around me and
make sure the team's got it. Obviously there's a bit more
around that I was looking after individuals, but.
But I, I think that middle management is going to be
radically impacted by it becauseyou can now have, you know, in
the system that I've described where you've got your own data,
(13:23):
you can put in your own performance information.
These are key objectives for thenext year that's broken down by
quarter. This is by team, this is by
individual. Then you can have a system
that's listening to every meeting, which we have, you
know, the doctors of the world that can automatically then go,
OK, how are you? How is this team tracking
against their key objectives? The manager no longer needs to
(13:44):
do endless performance and they hate it.
They hate it, yeah. But that the machine can put all
of that together and say here's how we're tracking and then you
can have the conversation that you need to around, OK, how do
we course correct? These things have have now
emerged. We're struggling in this area.
Like what do we need to do? So those kinds of things, I
think are just going to make people way better if they use
the tools. Well, if they don't then.
(14:05):
Well, I think it's the same as everything, right?
Garbage in, garbage out. And I think you mentioned data
before. The risk is for all these people
that have data, how good is thator relevant is that data or how
clean is it? And when I say clean, I mean,
you know, like did that? Has that customer left them?
You know, are they going to get put into the mix?
Because my worry is that they'regoing to Chuck all of that data
into some bot and they're going to do something to stop plowing
(14:27):
out stuff to people that don't even love them anymore or don't
want to hear from them. So there's a bit of that going
on, but I think that's the same as any use of data, right?
You need to clean up. I need to talk about omniscient.
Because I this is a good take away.
Yeah, because I, because I deliberately got excited about
AI because I am really invested in it and wants to know what's
going on. Tell me a little bit about what
it does and how it represents. And obviously it's quite closely
(14:48):
linked to what we just spoke about a data piece anyway.
That's right, yeah. And and at the core of it is.
A That's why I did it, JP. I was deliberately, I
deliberately LED, LED you down apath.
Thank you. Thank you.
Not well. Done.
So what Omniscient does, It's a data collaboration platform that
allows two or more parties to share their data.
And I use share in inverted commas because core to the
(15:10):
platform is that nobody ever gives their data to somebody
else. Yes.
But the kind of core use case that we were born out of was
allowing a bank to build a credit model from grocery data.
So where you know, I'm just moved here.
The Experian or TransUnion know nothing about me.
The bank won't give me credit. Not even a telco will give me a
(15:32):
cell phone because they don't they don't know who.
I am. You've got no background.
And it takes a long time to build up that credit profile,
few years. I think it was about three years
before I could get a cell phone contract.
JP was quiet. Yeah, exactly.
Where's your grocery data? That's very, very and recent and
I suppose highly predictable. Yeah, I was going to say
(15:54):
financial capacity. A lot of patterns emerge quite
quickly. Yeah.
And what we've learned, so the kind of use case I understand
was born in South Africa and expanded here, headquartered in
the UK now. And what we learned in South
Africa is that that grocery datais highly predictive of
financial capacity. And so there where they were,
(16:14):
the big retailer that we work with is called ShopRite.
They had 8 million customers that the banks didn't know who
they were. No, no credit information about
them using their grocery data, they were able to see, hang on,
three million of these should bein the financial system.
Wow, they can afford credit. It's not going to be bad for
them. They can kind of get into their
(16:35):
foot into the financial, you know, whether it's buying a
house or getting a loan for a car, like all of these things
help people progress in their lives.
But if you have no credit score,you don't have access to that
system. It's crazy.
And so this, this data collaboration enabled the bank
to appropriately and responsiblylend to customers that could
afford it. For the customer, they suddenly
(16:56):
like, wow, I can now when I've been denied before I've applied
for a loan to buy a house, the bank said no, because they don't
know who I am. Now I can use my grocery data to
do that. So that that's kind of the the
hero use case, if you like, that, I understand was born out
of as that world has has evolved.
So the platform has become more and more useful to allow maybe
(17:18):
not non endemic brands to collaborate.
So in the hospitality space, youwould have a car rental company
collaborating with a, a hotel. You know, that's, that's an
obvious connection. What you wouldn't have is an
insurer collaborating with a hotel to say or with the the
airline to say let's provide travel insurance.
(17:38):
And I saw I don't know this total side but but on Dragon's
Den there's ATVI don't know if you know the TV show Dragon's
Den, there was a young 2 girls on there who'd built something
for rental. Same story people couldn't rent
properties in London because they had no background and they
had no credit history and they were building exactly same
thing. So they were building trying to
(17:59):
build a credit history based on their other spending to allow
them. So a very similar story, I think
to allow them access to that. It got Pooh poohed.
There was something in it that they were saying that landlords
wouldn't buy into it or whatever.
It was a company it was but but it was a very similar story and
and just out of interest on on the South African side of things
did is that still is that still going?
Is that something that. Yeah.
So they they. Use that.
(18:20):
So in South Africa now most of the retailers on the platform,
all of the banks, all of the insurers, medical aid companies,
health companies as the and thisis where the kind of beautiful,
the beauty of this ecosystem comes into sort of a league of
its own. We have a concept of a data
seller. So that would be the retailer in
our conversation, it might be the restaurant or the hotel,
(18:43):
somebody who has customer data and says we think this is
valuable to somebody, make it work.
What the platform allows them todo is to put it into the
platform in a completely anonymous way, so they encrypt
any customer data before it leaves their environment.
So even if somebody hacked Omniscient, they would never get
access to their customer data, never left.
That's why I use share and enriched commas because it's
(19:05):
encrypted. So once it's in our platform,
even the grocer doesn't know who's who that customer is.
But the bank or the insurer can look at that data and put their
own dates in and do a distinct match at a customer level.
But nobody knows who the customer is because it's all
encrypted. So the bank and then look at
that. So let's say a restaurant had
their dates in there and the bank is saying we want, we've
(19:28):
got a, a credit card offer for high net worth individuals.
As an example, they could look at the customer data from the
restaurant loyalty program and DDupit say, well, actually these
10,000 customers are already ours.
But actually there are another. If we look at these 10,000
customers that we've got, these 2000 are good customers to us.
(19:49):
Who else behaves like these two,these 2000 customers at the at
the restaurant and they can thengo to the restaurant with an
offer to say, we'd like to partner with you.
We're prepared to pay you X to do a combined offer to your
customers. And once they, once all parties
agree, then the restaurant can then take that offer to their
customers. But they're always in control to
(20:10):
say, is this in the best interest of our customers?
Do we have the right to use their data in this way?
You know all of those things they.
But the anomalous part makes it so unique, doesn't it?
Because the fact that you can goalong and say, look, here is my
bunch of data. This is what it, this is what
we've got. And because the fact is they're
sitting on the state that has value that they probably don't
know about. That's right.
Or they're not thinking about the value that it has maybe.
(20:32):
By the way, it's not their business.
Yeah, and it's similar for the retailer.
The, you know, a grocery retailer's not thinking about
why is this our data valuable toa bank or an insurer and what's
the Amnesium platform allows them to do is to at no human
cost to them, at no risk to their customers and at no
expense put their data into a platform.
(20:52):
It's a no brainer. I think you might have the best
job ever. The.
Bank and then their data scientists look at this data and
say and we and that's the role of Nissan plays is to bring both
the buyer and the seller together.
Almost like the broker. Yeah, Data broker type.
Thing and in that environment have the we call it a you know a
machine learning environment Jupiter notebooks as the
(21:12):
platform where a data scientist can then build quite a
complicated look alike model if they want to which which takes
the credit information from the bank or the insurer overlays
that with the restaurant or the retailer or the grocer or
whoever they buy the seller is and then look at OK what kind of
offers can we come up with out of this data they can never
(21:33):
action it all they can do is go to the data seller and say we'd
like to do this because you've got yeah once the once they
agree then it's then they can you know and.
I probably should bring this in.What?
What do the commercials look like?
I mean, what potentially, you know, what kind of offer would
you think of with a card? Provider, It depends very much
on what that offer is, so if it's a mortgage matter, they'd
(21:58):
probably be prepared to pay £1000 to bring you a mortgage.
Customer, If it's a credit card,maybe a hundred 200 lbs who
knows. I told is this like data mining?
But someone was talking about they do, they've got a platform
where they're data mining and they were offering to
restaurants and maybe, I don't know if I understood him
properly, but they're offering to restaurants.
So sort of the similar thing saying look, you've got a bunch
(22:20):
of data you're sitting on, give it to us, we'll mine it,
whatever that means. And then we'll unearth diamonds
as it would be, or unearth quality for.
You value for you. You could use that term.
I think the the key thing in theAmnestian platform is they never
give up control, right? So the danger with giving your
data to somebody that's going tomine it is you relinquish your
(22:41):
IP. Yeah.
So someone, I mean number regardless of what they say,
someone could quite easily. Yeah, it could be or they find
value and then they can action it.
Yeah. Where once you've encrypted it
so you know the different degrees of you sharing or
collaborating or mining that data.
A key thing is that encryption layer where you've never given
up your IP, you've never given up access and you've never given
(23:03):
up control. And at any stage they can remove
any of those to say we we no longer comfortable with this.
Partnership and and is there is there a common theme when
they're doing when not just necessary hostility, but is
there a common theme when peopleare owning their IP and giving
up the data giving up something the right word but sharing yeah
I've got my an invasive camera yeah sharing their data.
(23:25):
Is there any kind of commonalitythat that that comes out of that
when they discover what they've shared, that they they're like,
oh, wow, that wow moment. They're like, OK, we've actually
got some really useful data herethat could absolutely.
Yeah, yeah. Is it very rare that you don't
get that? If if you don't have the
counterparty, So when the kind of ecosystems in its early
stages have you, you need buyersand sellers, yes.
(23:46):
And if you only have one buyer and they're not interested,
then. That's very true.
That's a crap deal. Yeah, exactly.
But but it the the the other like lens to look at it through
is for a restaurant or a hotel, they would have their own bank
that they bank with. So that that would be another
area where they can collaborate where currently you can't, you
(24:09):
can't upload your customer data to the bank and say who else at
the bank looks like our great customers.
Let's do a partnership with our own retail bank.
Just say to put an offer to other retail bank customers to
say come to our restaurant, cometo our hotel, Let's let's do
some. Is there like a bit, is this
quite a lot of an education piece then for the people's data
(24:30):
you want to share because it's because.
That's a bit a big part of a bigpart of the job is just talking
about the art of the possible and how everybody knows that
data is valuable, everybody's nervous about giving it away.
And so bringing those two together.
So I think education is key. You know, it's why we engage the
regulator. For the regulator, this is
critical to have the privacy of the individual protected while
(24:53):
not stifling the economy. The number one is their flag in
the sand, isn't it? Yeah.
And that's what omniscient does.It brings those two things
together to say we're not going to compromise somebody's
privacy. That's that's sacrosanct.
But we also want to create a data economy where businesses
can thrive. Yeah, it's having a responsible
platform to do that. And.
It does seem crazy, doesn't it, that we sit on all this data and
(25:13):
then it doesn't people see value.
I mean, sometimes in hostility, most people talk about CRM data
as e-mail marketing or sending something out on your birthday
or whatever it might be to get areturn.
But actually this is a very different conversation.
This is about the demographic ofyour business or the people
there's because there's different types of data.
Am I psychographic? Am I right in saying there's
different types of data that youcan get out of that, all
(25:35):
different types of learning you can get out of it and I think
understanding those are really key for everything, right, for
marketing purposes or what? You might want to do exactly.
I mean, if you, if you're running a restaurant, how
valuable would it be to go to your bank and know that actually
you've got customers who might live close to you or live far
away, That doesn't matter. But it would be great to know
(25:57):
that. Do your customers also go to
other restaurants? Yeah, What types of restaurants
do they go to? Where else do they spend their
money in an anonymous way? These are this is easy to share
and collaborate. Absolute gold.
Yeah, because. The bank's got that information.
Yeah. There's no mechanism to to
easily kind of reach. Well, I haven't.
Heard of you before. Reach that insight if you like
(26:18):
both parties. So while initially there's kind
of a clear seller and buyer in the relationship, quite often
those roles get reversed. We're actually now the
retailer's interested in the bank's data just as much as the
bank's interested in the in the retail.
And you're right, you'd never, why would you ever partner with
your bank? I mean, I just can't think of
any apart from to get a loan or and that's not really a
partnership, is it? It's kind of a one way
(26:40):
partnership that but yeah, but there's no other reason why you
would. And I think that again, it'll
build some structure around thatrelationship as well as, you
know, find out some true value in in what you've got that
you're already sitting on and what kind of numbers you look,
because I think they're sitting on a lot of data restaurants.
I think, I mean on average, I mean someone was talking to me
the other day and we're talking hundreds of thousands of emails
(27:01):
and whatever or customers that they've had through the doors
or. Yeah, yeah, yeah, two different.
I mean the obviously the coffee chains like the Neuros,
Starbucks, those are in the millions of customers, a
restaurant chain with 100 sites,they're probably sitting on a
few 100,000 customers that have depending if they have a loyalty
program or not. Obviously if they don't, they
(27:21):
won't know who those customers are.
But then again, if you've got card data you could part with
your bank, there are ways in which they can look at how to
kick start that sort of data economy that they can benefit
from and then you're appropriately.
Partnering and your AI knowledgeor your expertise in AI you'll
then because that's actually built into omnisci and obviously
(27:42):
the in the middle that matched the matching part is done by AI
presume or the yeah. We've got a smart machine
learning system that will look at if one customer's got e-mail
address, first name, post code, somebody else has got cell phone
post code and some other piece of data.
There are different ways that you can match that data to try
(28:02):
and look at, you know, what is the intersection of these
customers, but that's, you know,we'll be getting into a lot of
technology down that path and. Then, and you go through that
posters then and obviously there's some value to the data
that the customer is now aware of.
And the next step is that they deal with the partner, they deal
with the bank, whoever it might be, they do some kind of deal
(28:23):
and off they go. Yeah.
The the process is customer data.
Seller puts the data into the platform.
Yeah, we would then have our data scientists look at that
data and see what where do we think the value here is?
And what's that? Is that a long?
I mean, I hate to put a timeline.
Are they looking at a long period of time?
How long does that take them from meeting you and giving you
my sharing my data? I've got to find.
(28:44):
You've got to find a new word for sharing.
A new word? Yeah, Lending depending on
depending on who it is, if it's if it's a tier one bank or
retailer, they take a long time getting the data in a simple
getting over privacy and security is.
So much a bit more difficult. Yeah, yeah.
Because they want to know is it safe?
All all the right questions. But we've, you know, we've done
(29:05):
this globally with TransUnion, which is probably the hardest
way anybody can get through to have a Bureau that's prepared
to, to collaborate with their data.
So we know we can answer all those questions.
It's just a question of, you know, going or going through the
jumping through the hoops. Once you through that then it's
quite quick, you know getting data in as a few weeks.
And how is it delivered back to the customer and what, what kind
(29:27):
of how is it presented back to the customer in terms of what
you've found out? Do you know what I mean?
So as in I'm the MD of Chris's restaurant chain.
You've looked at my data. Do I get back some kind of
report that says? There would be a report, but
more often than not we would come with a proposition to say
we think this ensures. Right.
OK. So you would start that matching
(29:48):
process, right? OK.
That that's so omniscient. There are three things that make
make us unique when you put themtogether. 1 is that privacy
piece. Yeah, the, the, the
cryptographic encryption. 2 is the matchmaking bit that you're
talking about now. And #3 is that machine learning
platform that allows the buyer to create a quite a
(30:09):
sophisticated model out of the seller's data and look at the
intersection of them. So that works.
That was talking about. That is pretty much not done,
but by the time you've gone through the process of looking
at the data, you've done the matching or proposed the
matching. Yeah.
And and quite often we'll have another customer that's got a
similar data set that we've worked with before.
Yeah, I imagine that's quite common.
Yeah, imagine. You know, OK, this kind of
(30:30):
frequency of transaction with these sorts of purchases.
That could work with. Could be interesting to.
Travel insurance or pet insurance or credit card or
debit or you know what, whateverthe and.
What was it? Was it wasn't born out of
hostility that it was born out of?
Born out of the credit space, right?
That's right. The founders come out of the
more, more the banking side. So I think that that was the
missing pieces. Retailers for a long time have
(30:52):
been selling their data to theirsuppliers.
Yes. You know, the unilevers of the
world. Yeah, exactly.
Yeah. What they weren't doing is, is
selling that data to the banks. Yeah, but when you come from the
bank world, the banks for many, many years been trying to get
their hands on grocery data because they know it's so.
It's so rich. Yeah, it's resilient, I suppose,
(31:14):
because it is the most consistent thing a human will
do, right? Exactly.
It's very current, yeah. You know, right now the case
somebody started buying eggs. Therefore, the, you know, we've
all heard the stories about the,you know, when the retailer knew
before the dad did that the daughter was pregnant because of
what she was buying. Yeah.
And we've all heard those stories.
But that that's true. You know those those things are.
(31:36):
That's the, I think that's the really fascinating part about
it, how you can look at moments in people's lives potentially
without knowing the person, without knowing the moment,
yeah, if you know what I mean, without actually knowing what
happened and see changes, changes in spending, Yeah.
And changes in behaviour, which I think is so crucial for.
That's right. So the, the retailers have known
this, you know, Tesco's been doing this for 20 years.
(31:59):
There's no ways they would ever give that data to to a bank or
an. Insurance.
That's a chance, no? They'd be compromised by the
customers, you know, the integrity they have in the
relationship that they've built for 20 years, the.
Most successful loyalty program on the planet.
And so having a way in which they can let a bank look at it
without ever giving anything away and the bank can come and
say this is what we want to do for your customer.
(32:19):
And the, and the retailer says, is this in the best interest of
our customer? If they think, yes, it is, but
hey, let's let's collaborate. If they think no, they say,
sorry, we're not, we're not interested in in that, whatever
the use case is. And we've had many of those
examples where you know, the bank finds value or the insurer
does and. The side says you know if.
(32:40):
You know, if somebody's going todefault on their credit card
long before they default, if they're spending patterns change
or if they start buying lots of alcohol, you know, down the
line, there's going to be an insurance claim.
Yeah, But that's not in the bestinterest of the customer, No.
So the retailer would say, hang on, like while that we know that
data can be used for that, we don't want it to be used for
that. It's.
Very cool. And and and this is all insulin
(33:03):
I suppose with when we look at AI and everything we talked
about at the start of the podcast.
And I am interested to see whereyou think that might go in terms
of because as you mentioned, there's lots of GBTS and chat
GBTS or whatever, different versions coming out everywhere.
Where, where do you think it will, where do you think we'll
end up and how quickly will we end up there in terms of AI and
(33:23):
what and we talked about a little bit at the start, not
just in hospitality, but in general day-to-day use.
Do you think because I feel likeeverybody's kind of on ChatGPT,
maybe I'm not right. Actually my, my wife isn't, but
you know, but I feel like a lot of my work friends are all using
it now and kind of use it regularly.
Where do you think it might end up?
I mean, on the AI side, I think the, I think we're going to go
(33:44):
through a few waves. I think the next one is how do
teams work together? We've got individuals working,
but it hasn't yet gotten to a stage where companies have a
united kind of way of dealing with.
Like an interface where we're all.
Yeah, together. Yeah.
So that that would be, I think the next one, like a team of
engineers is a good example you have and individuals figuring
out how this makes them better. You don't yet.
(34:07):
We're not yet at the stage whereteams are looking at how it
makes that whole team more effective, so I think that
that's probably the next. Yeah, that's where I actually
had an instant the other day when we were talking about
something between our team and we were trying to take some
data, weirdly, maybe I should give you a ring.
We were taking about 60,000 pieces of data and trying to, we
were trying to work it through ChatGPT and chunks to see if it
(34:27):
could do much with it. And it was quite interesting
actually. It got, but then we were like,
we weren't using the front end of the data we were using there,
the systems they use the tech systems.
So we're trying to get some pattern out of it, but it kept
breaking. Yeah.
So so this is a classic example where having your own private
large language model that is trained on your data can
reference your data. Exactly that.
(34:48):
That, and to me, that's the nextstep.
Every company is going to get there from the smallest to the
largest. Yeah.
Having a private space where their own, you know, their own
machines, whether that's in the cloud or on their own hardware
doesn't matter. Needs to hurry up.
I mean, I, I genuinely believe that's one of the single such an
unlocking moment for a lot of companies because I think a lot
of companies just especially in hostility, we just kind of
(35:10):
bounce through day-to-day and wekind of just get there, get
there. And then to have that power of
understanding why we're doing what we're doing, what results
we're having, and kind of where the successes and failures are
or where the opportunities are without having to do too much
Excel work would be absolutely groundbreaking.
You know, and and there, once you've got those together, then
(35:31):
the number of use cases you can build on top of that is endless.
Yeah, you know from a simple, just your own private ChatGPT
interface that you can query. And then?
Tell me about this to a chat botthat sits on your website that
can answer customer specific questions or company specific
questions. To the performance management
one we spoke about that middle managers can use to track how
(35:54):
people are doing and have the right conversations.
It's exciting. I mean, I think it's brilliant.
I mean, I think I've only reallygot involved.
I worked with a guy called Patrick who's a large language
model expert, like a grad from Oxford, super smart chap, a lot
like yourself. And he and he, he talks a lot
about, you know, where it's going and what he's doing.
And he's built some incredible stuff with it already.
(36:15):
And I, yeah, I think it's an opportunity that everyone will
be is already grasping, but we'll probably be there quite
quickly. Look, we only get half an hour
with you people and you're superinteresting.
What's the best way to get hold of you?
If I can, if I can send in my listeners your way.
Do you have time for two? On the omniscient side, it's
omniscient.com, right? And for the AI consulting, it's
adaga.com. Adaga, very nice.
(36:36):
Is that your, is that your business?
That's my business. That's your consulting business.
Very cool. Look, Thank you very much for
coming in. Enjoy the sun.
We've actually got 24°. That is something.
It's beautiful. Yeah, well, I wouldn't know.
I'm in a dark room with their cups on the wall.
So that was that was JP everybody.
And we shall see you next week. Thanks, JP.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
New Heights with Jason & Travis Kelce
Football’s funniest family duo — Jason Kelce of the Philadelphia Eagles and Travis Kelce of the Kansas City Chiefs — team up to provide next-level access to life in the league as it unfolds. The two brothers and Super Bowl champions drop weekly insights about the weekly slate of games and share their INSIDE perspectives on trending NFL news and sports headlines. They also endlessly rag on each other as brothers do, chat the latest in pop culture and welcome some very popular and well-known friends to chat with them. Check out new episodes every Wednesday. Follow New Heights on the Wondery App, YouTube or wherever you get your podcasts. You can listen to new episodes early and ad-free, and get exclusive content on Wondery+. Join Wondery+ in the Wondery App, Apple Podcasts or Spotify. And join our new membership for a unique fan experience by going to the New Heights YouTube channel now!