In the second episode of PaymentGenes’ “Voices in Payments” podcast series season 2, Sjoerd Leemhuis, CEO & Co-founder at Owlin joined our host Diederik Klopper. Together, they dive into the use of AI in detecting risks and trends. This episode is a must listen for any company looking to educate themselves about using natural language processing and AI to turn data into comprehensive insights.
Owlin develops proprietary, next generation, Natural Language Processing (NLP) technologies providing organizations with critical insights into counterparties, suppliers, competitors relationships and prospects proactively, continuously and in real-time
Listen to the podcast to find out:
If you’re curious to know more about Natural Language Processing or Risk monitoring, feel free to reach out to Sjoerd Leemhuis for a conversation.
About Sjoerd Leemhuis
Sjoerd has many years of consulting experience in both Banking and Payments in The Netherlands. While seeing many corporates from the inside, he came to the conclusion that both worlds would be better off with a solution such as Owlins'. He combines consulting with a deep understanding of NLP, Machine Learning and how the two are best combined.
About PaymentGenes's "Voices In Payments" - The Future of Payments podcast:
The “Voices in Payments” Podcast, is an initiative launched by PaymentGenes to positively impact the payments community, by educating and connecting the market with vertical-specific industry expertise.
Growth by providing expertise-driven Recruitment, Contracting, Greenfield Consultancy. These services all resolve and intersect around payments. Learn more about how we can help your business here.
00:00:00
Hi everybody. Welcome to the PaymentGenes podcast series voices in payments, and welcome to season two, where we talk all about a, B, C, D of payments. In this modern day and age, there's so much information available that it's very difficult to see through the relevant data. My guest today is Sjoerd Leemhuis, and he's the co-founder and CEO of Owlin. With him, we'll talk about how to use natural language processing and artificial intelligence to make sure you have valuable insights.
00:00:36
Okay.
00:00:37
The founder of Olin, a very warm welcome to the show.
00:00:40
Thank you. I'm honored to be here. How are you doing today? yeah, I'm doing pretty good. And given all circumstances I'm happy. Did I have a good reason to leave my house?
00:00:49
Okay. It's good to finally have a face-to-face meeting again. It's been some time. Yeah,
00:00:53
We, we keep distance to have a face-to-face meeting. It's great to have you on. I always start my postcards the same way,
00:01:01
You know, not an hour. How did you stumble into payments?
00:01:05
Ooh, I, I liked the question. Before I, started out in as a company, I, actually works as consultants in the financial sector. So it works within ABN AMRO bank. I also work with the rival bank and, within the rival bank, I actually was responsible for integrating, credits cards, products in cards, acquiring in the. Some people might know that one. And, for me that was the first step into a wall of payments. And, then a whole new world opened up new worlds opened up. We had to create a business case with tenders to advise other people to integrate it's credit card acquiring. And, and I remember one nice anecdote is I, first time I met, the guy within the organization with the bank who knew everything about the schemes of MasterCard and visa. When I looked at these desks, he had all the schemes printed out on his desk.
00:02:00
I think, yeah, I think there was to start with big data. I think that's how I rolled into this space, as a consultant in the first time and later I found out it was no surprise that we also still have a lot of, clients in the payment center.
00:02:18
Yeah, exactly. You started the own and coming from the financial sector, what was it that you thought, okay, this is something we can do better.
00:02:27
Yeah. So, so what you see in, one was a big problem today is what you also could see already, within the payments world is, there is more and more pressure that you need to keep on track of everything what's happening. I made this joke about the schemes from Oscar to visa, but it's actually day dictates, Hey, you need to be aware of what your merchants are doing and you need to be, otherwise we will kick you out and kick you out. Please check what you're not committing fraud, where you're not committing, criminal behavior. Also you want to check whether they're financially healthy with everything. And, and I think we, what we saw was, okay, there's a lot of information, yeah, overload and it's getting more and more information into these worlds today. It's hard to, identify the relevance, signal from the noise and, what we are a very good debt without it is.
00:03:23
We look at the news all over the world in all different types of languages. We, analyze that with a big natural language processing platform. We use all sorts of tools to distinguish the signal from the most. We help you to identify what's going on in your portfolio. It's going to be also in your portfolio of merchants where we say, okay, this is why we see certain, declining revenue, or here's where we see, people being, brought to courts or here's where we actually see maybe even, profit warnings, fraudulent behavior, or a lot of other types of bankruptcy signals that we'll pick up as well. This is, was really hard to do this initially, because as a human being, it's impossible to be on top of a portfolio of thousands of companies. It's just not doable. You don't know where to start. You cannot read all the information every day, and that's where technology helps a lot.
00:04:17
I think that's how we saw, Hey, there's a lot of, there's a big problem out there. People are trying to solve this by throwing humans at, it's.
00:04:28
Not a device there that's better equipped for it.
00:04:29
That's the start off without in how we are we getting to this point where we started to analyze all the news out there, all the techs out there and translate it, analyze it, and bring it to people who are in needs of staying on top of the game. Yeah.
00:04:43
You mentioned a few interesting topics. You'd need multiple languages, natural language processing, but first of all, how do you select, for instance, which languages to, take into consideration first, or how do you define the scope there?
00:04:59
Yeah, that's nice. I think, what we have seen also is I think if you take a look at the, what I would call the old world, you would have, everyone was reading maybe one newspaper and in the Netherlands, she has the finished edit lots of fish.
00:05:16
If you're a bank or a bank.
00:05:18
Read, of course the data tab for the NSA, but a lot of information, and if you take someone global scale, it might be like Bloomberg or Reuters release the data that's in the rules or leads to the information to people again. If you take a look at the world of, payments, what's, awesome situation is a lot of merchants are small, medium enterprises, a lot of small, medium enterprises.
00:05:42
They were local market.
00:05:44
Local markets, and that were, are in certain niches. What you see there is, a lot of information that's very relevant and never gets rules to mainstream media. And, it lives more in local news outlets. And, and we specialize in analyzing all those local news outlets. For us, it made a lot of sense. Also, we took a look at our clients and, what you see is a lot of people are investing in emerging markets, or China is booming. For us, it made a lot of sense also to start looking into Chinese news, for instance, Chinese articles. We see the same in Eastern Europe. We see the same in, in South America. We support a lot of respect to Russian, a lot of European languages, but also Spanish Portuguese. That's how we, not only started to translate those languages, but also made sure that we track a lot of local news sources in different countries, in order to be complete, in order to be in the know and be aware of what's going on there.
00:06:48
Yeah, here's, I think in this day and age, it's quite easy and I think, recent news items have shown that it's easy to get into a news bubble that had that confirms in the things that you already know and validates your point of view. Of course, I think the proposition that you have, that he will provide a three 60 overview of the perception of civic, companies, how do you then select which, news outlets which media companies to include or to put on a, on the sidetrack?
00:07:18
Yeah. Yeah, I think what's, I think the news bubble is extremely dangerous to government, and I think if you are a professional, you never wants to be in that bubble. And, and it's good to also realize that, news bubbles are created also by social media platforms, in order to keep for advertising purposes rights and to get people interested as into be a stay on the platform. We serve a lot of risk managers. We have a different perspective, completely different, so you definitely want to step outside of the bubble. Also wants it be aware what's, what extends might I be in a certain bubble, and make sure that we also focus on the stuff that's happening outside of the environment. We do not discriminate on sources that much, or we also do not value other sources more than others. We do look at, we have a closed system also of sources that we track.
00:08:17
So, I will give you an example if, we do, they can start a new website and we can put Twitter and Facebook share buttons on our websites, that immediately we'll bring our websites into the news flow or Facebook. Our system doesn't work like that. We, it's eventually also analyst and needs to say, okay, are we really going straight to the source? Yes or no.
00:08:40
First to check whether the source is credible and there's,
00:08:43
There needs to be certain credibility. What are sources, maybe political flavors that you still find, because you still want to be aware of.
00:08:52
That's of course, the shapes general,
00:08:56
Satiric websites, for instance, you don't want to follow so dispel, it is out of the scope depth. One is out of the scope, you have the onion there and stuff like that. The spells in the Netherlands, that sources we don't trick. A lot of others we do, and also to make sure that we, yeah, keep people. It's a lot of information. So it's also, the case. It's not that we felt well that the information to people, we really pre-process all the articles. We have a system that's, reasonably articles, that's classifies, which companies are mentioned here, which people are mentioned here, which locations are mentioned here. Also, what is this about? This is about a board change. This about a bankruptcy signal? Is this about shops Logix or profits warning? So we have, we classify everything and then we built a whole visualization that tells you within your portfolio.
00:09:45
This is what we see when it comes down to bankruptcy risk. This is what we see when it comes down to board changes. We quickly completely turn it upside down, that is actually.
00:09:56
More proactive than reactive, way more.
00:09:58
Proactive. You, as a user, you don't have to read all the millions of articles a day. We tell you, this is where we see direction. You might want to look into these, versions. Yeah.
00:10:12
This is, are you familiar with it? They have a magazine called three 60 magazine. Yeah. Yeah. It, perhaps for the listeners who don't are not familiar with it provides, different news inputs from surrounding one specific topic from all over the globe to really get a three 60 view. This is basically what you're doing for companies. Right.
00:10:32
It's very comparable, I would say. It's really, we also follow a news stories if they, and if they develop over time as well. It was always easy to, if you would look up a specific company, we made the, give a timeline, where you can see what happens with it.
00:10:52
Oh, so in the passage you have the data.
00:10:55
No, because we have the data and that allows you to browse and get a more holistic view of what's happening.
00:11:02
How far back then do you go, did you start building the database when you first started, or did you also scan all the previous articles that were written in.
00:11:10
That there are some abilities to also go back in time? but also, what we also have seen is that news as a tendency also to, sometimes disappear or be archives, or sometimes even, news articles get re edited, we track all these developments. I think we can say that's a, that we have, more than a five-year of history, about seven today. And, but we also see if articles have been changed over time. Even though we can also see if, articles are, have a wrong at timestamp, because that happens as well. That's, that articles sometimes are being backdated authority. I can republish.
00:11:54
So years ago and that's,
00:11:56
That's something we can also analyze and see on the internet.
00:12:00
That's cool. Let's go back a bit to the early days when you first started out in, of course you're building a product and you see a market fit, how did you get started in linking up with the first customers and what were the main use cases that really resonated then?
00:12:16
Yeah, I think, so I worked in the financial sector before you had the contents, also in payments. It's, it's very cool to say that I think, one of our very first clients was a payment service provider, being Ingenico, today we serve way more at Bay, miss years, providers, 40 social in the Netherlands, which are agendas as a client as well. And, but we also, had, ING bank, are still applied today, with, very comparable use cases. So, so often what we see it's hard to track a portfolio, of a lot of companies were very few people and,
00:13:01
Okay, bye. See,
00:13:04
Portfolio enrich monitoring, often what you see there is, there's a big pressure on, well, digitization, there's also a big punishments,
00:13:15
Very expensive and need. Yeah,
00:13:17
True. And, and there needs to be, more and more needs to be done, with less and less people. There's also an increasing pressure of, regulators, mostly the visa schemes and so on. That's where we flourish because then technology can really make a difference as you can do a lot with a few people. I think if you take more and look at use cases, for instance, where a lot of people are investigating a few stocks, for instance, then all the news analytics, it still provides some value. Then, but that's not our sweet spot. I think it's more big portfolios, a few people, yeah,
00:14:01
Excited, because I think then you really have to scope the pure width of the sources of information that are valuable, and there's no real way to process it other than through.
00:14:12
True, true, and one development that we also see, and that's actually that evolved over time. We also help, asset managers, by checking what's happening in the world. What we noticed there is, if you take a look at listed companies, that's, there's often a lot of information already available, so and all these, this information, but, there, you can only as moderately with analysis from local sources, if you take a look at the normalist universe, which, and more and also a lot of merchants that are need to be easy to be, attracts, there's not a lot of, tooling available to attract list of companies. I think that's where we, yeah, so a sweet spot, and that evolves a bit more over time while were growing the business. Now we notice, Hey, this is actually, we're really good to in tracking all these companies, let's focus more on lower cycle use cash.
00:15:20
That definitely makes sense. In this series, we're talking about the ABCD of payments, AI, blockchain cloud, and data. Well, I think there are two main topics that are of key importance to you, AI and data for a non AI understander, or for somebody who's not that technical. How do you differentiate different types of AI? Because if quite a lot of companies nowadays are saying, yeah, we use rent new AI tooling, but it's very difficult to really get a grip of what is actually happening behind the scenes. How do you think, companies can best validate how good in AI is that is defended by the company?
00:16:06
I think, when it's about AI, I think it's a space. The AI space is also a crowded by a lot of it specialists that are, not actively using AI using rule-based engineering, which can also be like rule-based algorithms. Right. So, which is all fine, which also can, you can still do a lot. I think when you use AI, you really have like feedback loops in forth, or you use training data to increase, the outcome of your algorithms. There's a bit more learning aspect in both. AI is not the solution to everything, because there's, a lot of things that you still want to keep in rule-based systems. I can give some examples because I think what you, what we often see is, we serve a lot of clients that are, that also are regulators. You want to look at what type of risks do I face in the markets, if you will, have a completely unsupervised AI model, saying, Hey, a DSR to accomplish, you need to look at, and no one can explain it anymore, what you're doing.
00:17:21
Also if you cannot explain it to senior leadership and you cannot explain it to the regulator, you get some issues. So, we solved the problem by taking the best of both worlds by looking at, okay, which problems can we solve with, using AI. How can we also leverage on rule-based systems? that's can work together. We managed to create a system that we can train, like when has the features of AI eventually can also train into a rule-based system, which can be on the streets and adaptive, and also explains to leadership.
00:17:55
If I answered correctly, the AI changes the rules that it's analyzing, so that you can have that as a baseline. Yeah,
00:18:02
That's correct. That's only for a specific part of the architecture, I think for identifying, companies, we use on supervisory, and that's because that does the trick for us and the 3d skills very well. I was really understanding your use guys first, and then on the standing, are we going to solve these with AI? Where are we going to solve this with a rule based? And I think we are experts in natural language processing for us. That's like the big umbrella, within natural language processing, you have different methodologies, and we always try to combine the best of both worlds to get to optimize outcome.
00:18:37
Okay. Explain for the audience a bit in more detail, what natural language.
00:18:41
Yeah. So, so basically, it's, what you try to aim for with natural language processing is to structure unstructured data. And, so it sounds maybe, but how can you do this? okay. In our case, we look at millions of news articles a day. We structured them by saying which company names have we seen here, which events have we seen here? how do they relate to each other? So we can also build graphs if this y'all relations, and then we create more and w this is what we call meta data, right? So it's a date out of the data, which you can then use in your investments or in your analytics or your risk processes. We structure it in a way that we can give you a metric, which has bankruptcy risk is increasing, over specific client of ours client's portfolio, or if you work with audio compared to peer app portfolio.
00:19:39
And, and how do we analyze that is by looking at all the articles by saying, what is this about, if we see that there's more gender about bankruptcy, we, that the score starts to increase, and that's, you can do that with natural language processing, by the computers analyze all the data. I think what people are really good at is read one news article and then really understands, that, the core message of the news article, computers are not entirely there yet. They are getting there, both, well, computers are extremely good at is, doing this at a 95% accuracy, but over cycles, within minutes. And, and that's impossible for you. So, so that's how you actually want to leverage the parts of technology, can already help you in pre-processing a lot of data and, help you getting the insights that you eventually need for your investment process.
00:20:41
What are, what are some of the most interesting insights that you were able to get out of your products that really took you by surprise? I can imagine there are a few things that came up saying,
00:20:53
Oh, that's a, I, there are so many surprises that we have, seen over time. And, yeah. Okay. Well, one of our fishes in the beginning was, when we leverage news from local news sources that didn't become mainstream media. Yes. You can also, out-compete, the markets in some cases, because you are vulnerable faster, you're quick on the ball. And, and what we have seen there is it's really interesting because we have so many cases in our system, where you, out-compete the markets, both with what you didn't see simultaneously is, it's not that easy because a lot of big news items, it's very hard to predict what are the markets will really move or enough based on that. And, and there was a big, what surprised me was how many guests we have, and how often this is still the case. There's a big information assymetry in the markets, both with also surprise me is that the market is not always moving, when things become mainstream.
00:22:03
I think that's maybe the biggest surprise. That's a, that we got, that's a, that you think while you almost think you have a Holy grill, but then you see yourself.
00:22:11
They're storming at the Capitol, but the stock exchange stays near any level or something like that,
00:22:15
Something like that. So, so that's really a, and then you are thinking, okay, there are way more variables, like, yeah. Of influence in, in, yeah. Yeah.
00:22:29
It also sounds like something that you could use a setting to traders because of course headed that market perception and then new stories that are around could also be interesting to them. It that as well, a client base of.
00:22:41
We, if you take a look at our client base it's famous companies, which is a big, client service virus, but also asset managers. Those are more portfolio, investment professionals, banks, consultants, and a growing group of traders as well. So definitely yes. And, and what you see also here is, it's the same, if you want to attract a big portfolio, but also if you want to attract companies in foreign languages, so what's happening in Russia, what's happening in China. There's not too many solutions out there yet that can do this. And, and I think what we started very early with, making sure that we are able to have those news sources, but also we are able to translate them and take them into our models. And, and I think that's, what you see with all sorts of asset managers and also traders approaching us. They're also, the interest is in, emerging markets.
00:23:44
I could imagine. You're currently processing 11 different languages, right? Yeah. I think it's 12, already 12 already. We're moving on 13 or 14 is increasing. So how do you now choose, okay. Which, which one is next on the horizon and how do you go about implementing that? And it's because there are so many variables and it's being able to translate from one language to another, but this as well, making sure that you check all the media sources and that.
00:24:12
Correct. Correct. No, shall we have team also responsible for the quality of sources, and we have also in our data science team, we have a few experts that know all about machine translation and, we are big consumers, also of cloud technology. So, we run Parsi, Amazon clouds. We run bars in Google glass, and we got, also approached by both parties to, threw together a good deal on translation. For instance, we projected that, let's see what we made some Galatians, will seek their own software and numbers. To the newest particles that we try and say today, that that's incredible amounts. That's where else paid off. Do we invest in machine translation? also, what we have seen is, Google States is drains to be a very generic translator. You have a specific industry focus, we focus more on that financial or boards related news. And, and that's that we see by training our models and such that we perform better in some languages.
00:25:25
And, then the Pierce,
00:25:28
Which of course is interesting for instance of Google to make, if they are able to translate on your behalf, they will, their systems will get better as well over time. So there's a truth.
00:25:39
I think what'd you, what you see in the whole machine translation space right now is that the models become more specific. If you are in healthcare and you want to train specific healthcare and machine translation models, if you are in finance, you want to, translate more financial, yeah. Specific machine.
00:25:56
Transactions, if the industry jargon, and.
00:26:00
It's the same for humans rights and deviations mean payments, as well as people love using deviations, both buts, but they might mean something completely different outside of the payments rules. That's a, so you need, so-called text matters also for machine translation, which are all AI based.
00:26:19
It's funny, you mentioned it because, natural language processing, neuro linguistic programming in mind when I first read the three letters. So yeah,
00:26:28
It's, there are some similarities maybe almost right. It's well, there's not the same, so it's really a, you need to, but this, because context matters. I think that's what it connects maybe together. And, but that's true. That's about it. Also can mean different things in different worlds. That's the whole, this whole situation. Yeah.
00:26:49
Then, I saw on your website, I think it was a quote from, again that they started working with your software. We're really happy about it. I think they started in a KYC AML.
00:27:03
So, so what they, how I didn't started working with us was, monitoring their at emergence, also on, yeah, signals that happened outside of so-so as a payment service provider. You already have information about the payments data. That's a, and so you're on top of payments, but sometimes also things happen outside of the system, but you control a change of ownership or whatever ownership, or a board member gets through with, or there are, orders get canceled with what are maybe big bankruptcy signals that you don't immediately see in the Batemans yet. And, so there were a needs of, to, as can track the emergence, on bankruptcy signals on fraud signals, completely outside of the data that they already have themselves. That's how, that's how we started with budget. Yeah.
00:28:00
And, I saw as well into quote that's afterwards. They started using in different departments as well. What brought us the main, requests that you got from the market saying, Hey, listen, we want to make use of your software in other ways than we are currently doing.
00:28:14
Yeah. So, so if you ever, if your company is growing, it's a base is growing, you also gets, you start to get more different types of risk teams, deemed responsible for the merchants, but also on our team responsible for instance, for your suppliers, because they have an increasing number of suppliers, and then you also start to realize, Hey, I'm actually very, vulnerable to a couple of my suppliers. Are you also want to monitor what's happening? There is very similar to tracking your merchants. You just want to be aware how likely is it that they're still going to be there tomorrow? And, so that's how things start to evolve. If you track your merchants is it's more imposed from the schemes from mascaras and fusion that you would need to be.
00:29:05
Yep. And regulatory, institutions.
00:29:08
It's UC as well. Of course, once you start with dating a banking license, there are some whole bunch of things you need to track it. And, and that's also how it's outgrew. That's a, as you start to realize, Hey, we also need to check GDPR data breaches and so on. So there's a lot of additional, yeah. Check the book exercises that you need to meet, where we can help. And, and that's how it grew over time as well.
00:29:35
Yeah. And, of course these companies have a specific dashboard, so you already mentioned it before, could get to talk us a bit through how that works. How malleable is that and how do most companies set that up?
00:29:48
Yeah. So, so we have full pipeline of data, mining analytics, and also visualization. We, our dashboards makes use of our own data, feeds our API. What you also see is a lot of our clients use, our API directly into their own tooling that use. And, so to just use the extraction of all the information stream, use the scores and the insights and then they can integrate it in their own visuals, or the alerting structure and so on. Our dashboards, which is also, more than half of the, I think it's more or less 50. Half of our users are also desperate users. And, there you log in, first thing you see is your own environments. The first thing you see is a, it's a beautiful graph, and we love bill, charge. We almost, we lose if we could get a base until next to difficult, lots of people know us from the belt with jars and, where we visualize your portfolio on any given topics, or it can be all risk is going to be a general cheddar.
00:30:55
It can be maybe impacts of COVID-19, so you can select spots, but our default setting is let's take a look at the risks in your portfolio. We visualize in a bit, which are your merchants, or your portfolio, where stuff is happening. You can have thousands of companies who are tracking, you only see maybe 20 that are officialized that are most important because that's where something is happening and how do we do that? So we look at news volume, we look at the contents, we look at a couple of different barometers and, that's combined tells you, okay, these are the companies that you need to look at today. You can browse through that list of companies.
00:31:36
For instance, how often an article is shared or likes or posts, is it one of the things you notice as well? Because of course that plays a critical part in the general perception of such a news article.
00:31:48
Definitely, definitely. So, so we look at it, we call it the clustering. Wherever you say, Hey, this article gets traction in the news. We also know it's positive or negative, and Austin also know more specifics about it, lots of events. If that grows really big online, that also knows what in place, more impacts. You want to know that they're still the same story. We say, Hey, it's one story, but it's big. This is what you need to know of that merchant. And, that's how we, how it works. What we also do is take a view completely separated from your portfolio and just looking at, okay, what are the main things and main trends being discussed right now, and your hope or thought yo, and all of a sudden, that's also merchants are also related to individual review or this slightly different view, which gives it all additional contextual information for, yeah.
00:32:39
For, for risk or monitoring teams to understand what's going on.
00:32:43
Yeah. Perhaps a bit of a sidetrack question, but are you allowed an issue, do you make use of your own software when you're contemplating investing?
00:32:54
We, one of the downsides is if you work with asset managers as your clients and also with banks as the clients, one of the what's very important is, trust. We all, science agreements to get our, we're not using Alin for our own trading purposes. Also because, not in all cases, but in some cases we can also track what other people are following, right? You don't want to have any of those, information to be used against you. So, so we, I use Ali, I track my own clients for instance, I, I track my markets. I do a lot of things with Arlin. I don't make active trade decisions based on loudly, and I put everything in Symfony, and I'm just busy working and growing up.
00:33:52
Did you ever have to call a client saying, Hey, listen, things are not looking that well, I've got you on my dashboard that, Hey, there's a lot of chatter. What's going on?
00:34:02
We, yeah, it's okay. It's yes. That's about a German payment service provider that we also have.
00:34:20
Yeah. So, we'd all know what happened there.
00:34:25
I think we all know what's happened Derek, but it's good to know that, yeah, I think, still what you see, payment service providing is such yeah. The, the job they fulfill in the ecosystem is so critical, that you also can see, that it's almost impossible to go out of business because a, there you need to stay up in the air to keep the ecosystem up.
00:34:54
Yeah. I think as well, that specific example, it has not proven well for the entire industry. I think the entire trust in the industry has taken a blow. I know a lot of our clients have received calls from their clients saying, listen, where's the money stored. And, tell us a bit more about your financial position at the mountain. So, yeah,
00:35:16
Which I guess, and I think what I didn't like about, our, role in the ecosystem is that we, soft, information, cemetery, and we provide more transparency. Our mission is to shape a better informed Wolf. That's really how we also, highlights what's happening the world. So, so we are after exposing fraud cases, we are trying to.
00:35:41
Of course, vital tool in the already existing tools that are regarding KYC, AML, fraud, checks, whatever PSPS and bang. I think it's a perfect addition to what they've already been using to really increase the efficiency of that.
00:35:59
I think, I think what you also see is, we see a big increase in demands, also given to all COVID-19 situation, because I think in the past, a lot of companies always relies on existing ways of doing business and doing their work. I think, the, all the lockdowns and all the additional rules, and if we're in boast upon the economy, just because of the COVID-19, they have tremendous impacts. And, it's really hard to predict which companies will survive and which ones and I think we really help in a structure that's better. That's what this conversation in highlighting, which companies are hurting most and its most. And, and we have a lot of people also call this now casting that's, if you cannot forecast, you can always now cast, it's near real time. It's near real time. I think with the analytics that we do, I think in many cases you can actually do proper forecasting because very often you receive signals that's from, yeah.
00:37:10
From one event that we're built to the other, both, but you also see situations where travel companies who are, if you Asian companies are almost bank routes, but then still get a sports baggage from their States, that they will keep longer in the air. I think that's almost impossible to predict, by the VC is very good to now cast. Yeah.
00:37:29
Yeah, exactly. As well, if there's a general tendency across the globe that the support packages are given out, then perhaps had that could be forecasted for one that hasn't received it yet. This is not a given of course, but it's secretly in the market.
00:37:40
It was on, I think that the trickiest can you get more certainty into your forecast? And I think with, natural language processing or news with the shingles we provides, you definitely can. Yeah.
00:37:51
Of course a massive part of that is having more data. Of course, the longer that you keep doing this, the more data you have, what do you see in a future of additional possibilities of predicting or analyzing that we haven't gotten.
00:38:04
Right now? Yeah. We've talked a lot about natural language processing and also on natural language processing on or news for if we're to identify risk. Well, there's more data out there. It's not just news, which can feed this process. You see a big increase of, during the data landscape, which is growing, you see, goes from credit card payments data, to satellite images from Garth sparks at the Walmarts. Right. I think all these examples, a lot of people already are aware of this. And, I think what's happening right now is that it's more important to connect the dots. So, and it's actually not that for us, a very logical step is also to integrate with, was there more, known KYC providers ready to give financial scores and financial data, because she wants to get, if you want to wholesome holistic view, you want to know about financial health.
00:39:11
You want to know about, markets, insights. You want to know about specifics of the health immersions, but also start including more and more authentic data into this, into your insights to get a true holistic three 60 view.
00:39:26
Yeah. Yeah, I need, so there are still a lot to discover there, that we're looking at on the near horizon next two to three years, what developments can we actually materialize over that period?
00:39:40
Yeah, so if you very concrete, I think if you take a look at how we started with just Noosh, we opened up our platform for all these other data sources as well. And, our sweet spot is still in Metro batch processing. Both reduced, gives the ability to lock in more data sources, to, create a complete a picture. It's great. It will be a picture and also to make sure that's where the one stop shop, where you can see over here, you have your complete workflow just in one two, because that's also, you see a lot of, dashboarding happening right now in the market. What you don't want this, 50 different types of dashboards for all different types of use cases still, I believe, that always works better than a lot of times people still use email for everything maybe automated that's even better, and you start using some tooling that's where it starts, but then see, how can you integrate doing better? And so it was more integration of dueling integration between sites, and also, making sure that you can run several different models, to do even better predicting.
00:40:49
So, so that you're not only looking at the news, but also looking at, great credit default scores from industries from, appropriately to scores, a lot of different things. You can, you can combine, in one overview to get the full holistic there's a few, and I think that's happening as we speak.
00:41:06
Yeah, indeed. So, yeah, still, like I said, a lot left to discover a lot left to do. Yeah, I think it's really interesting. There anything that we haven't discussed yet that we think we should, this is definitely something we should touch upon?
00:41:24
No, I think we covered a lot about, I learned a lot. I know AI to natural language processing. Well, I, I think the main trends that we see right now is, I think in the past, AI natural language processing, dark beings were we're, in the innovation centers where people were running proof of concepts. I think what you see happening right now is, it becomes more integrated in actual work flush. And, you see also concrete use cases and problems being solved. Yeah, it's still early stage. I think the, what you see is that the early adopters are working with this and, they understand, and I think a lot of, other companies are catching up right now, is everyone on the statute need to be on the strain, because it is a future. With that, you also see a lot of additional, funding is going to AI initiatives, more doing gets available.
00:42:35
I think, that's, that makes it, more accessible for a lot of people. And, but also I see that as a big plus, because it's, it makes it more people are more aware of the capabilities. They can also appreciate.
00:42:54
Because you better understand it is what you're working with, and it was also easier to trust it,
00:42:58
Easier to trust teachers to work with. I think, in conversations that we have right now with new clients, in the past, we also had to do a lot of, well almost, teaching, or lot of explanation. And, and I think now, we see that there are very well-informed lions on the other side as well, which really, accelerates, also the adoption of.
00:43:24
Sales process, shorter, our issue.
00:43:26
Definitely make sales process shorter. And, and that's also what we have seen with the whole, COVID-19 situation. That's, people want to sense, okay, we need to change, because there, I mean, the real storm, is, but also the real big adoption of the eyes also to,
00:43:47
Yeah, exactly. To wrap it up, there's a lot more data out there and a lot more news, no different news outlets. Having that the best way to structure debts, to put it into one, language of which you can extract all the valuable data, that's really the way forward and that ha making sure that you can draw predictions from that data. That's key. I think that's,
00:44:12
That's definitely true. I think that's also, it comes all back to our mission. Again, I think we had to share a bedroom informed Wilson and we started with DEXA news, both. We also make, see, makes a lot of sense to put more data insights and graceful transparency and make everyone in the world to be aware of everything is happening. Yeah. So,
00:44:32
I think it's a good feature to look forward to. Sure. Thank you very much for being on this podcast. I really enjoyed it. Thanks for y'all honor.
00:44:39
Every day I enjoyed it too. And, let's keep each other in the loop and see what's happening. Yeah.
00:44:44
Oh, all that's happening with us, so I'll keep you in the loop as well. Thanks for that. Bye.