Enhancing decision accuracy, with Maik Taro Wehmeyer
"If you help a bank to create lending IP, there's always a question around 'who now understands how it works' and 'who owns it'. And the lending IP, that is something that, as a fintech, as a bank, you should own yourself.
And that is why (instead of consulting) we decided to build software that helps internal risk teams and internal credit teams to create more lending IP, faster, and on their own without having any external dependencies."
Data used to be scarce and rigid, and so the legacy systems we built were designed with that in mind. But data is now everywhere and fluid, and those rigid legacy systems are starting to hold us back. Taktile is a modern decision engine built on a local no-code platform that makes it really easy for credit and risk teams to build and run automated underwriting decisions with a really high level of accuracy.
Find Taktile at https://www.taktile.com/ or jump directly to their State of Lending report at https://www.taktile.com/state-of-lending-2023
As Maik mentions on here, both he (https://www.linkedin.com/in/maik-taro-wehmeyer) and Taktile (https://www.linkedin.com/company/taktile1/) are active on LinkedIn. I'm there, too, so feel free to send me a connection request while you're there: https://www.linkedin.com/in/brendanlegrange
My action-adventure novels are on Amazon, some versions even for free, and my work with ConfirmU and our gamified psychometric scores is at https://confirmu.com/ and on episode 24 of this very show https://www.howtolendmoneytostrangers.show/episodes/episode-24
If you have any feedback or questions, or if you would like to participate in the show, please feel free to reach out to me via the contact page on this site.
Keep well, Brendan
The full written transcript, with timestamps, is below:
Maik Wehmeyer 0:00
For me, the power of a modern decision engine like Taktile is that it enables teams to leverage a wide variety of data sources - whether it's internal data or whether it's external data - when they are building their underwriting decisions and their credit policies.
Brendan Le Grange 0:19
I'm not a monarchist. Staunchly not. In fact, when I was writing the script for this, I quickly Googled to see when King Charles was coronation was going to be, because I felt like it was coming up soon, only to find out that it happened months ago.
I'm sure he's fine. I just don't think we should elevate anyone to such levels based on nothing but the lack of their birth, because it props up a very problematic class system that continues in my mind to hold back the UK.
But that's not why I was researching the coronation. No is because when you type in carriage drawn by eight horses, the top return is invariably the gold stagecoach and eight horse drawn carriage that was first commissioned for the 1761 coronation of George III. And that has been used in the coronation of every British monarch since William IV. Because yeah, being pulled by eight horses is pretty damn kingly. But what if you, or rather your recently formed startup, was being backed by eight unicorn founders?
Because that's what Taktile has. Taktile allows companies of all sizes to build run and elevate automated decision flows without requiring developers to write complex code.
A few weeks ago, you heard co founder and CEO, Maik Taro Wehmeyer take me through the findings of their first annual State of Lending report, and now he's back to talk about the business itself. Welcome to How to Lend Money to Strangers with Brendan le Grange.
Maik Taro Wehmeyer, co-founder and CEO at Taktile, welcome back to How to Lend Money to Strangers.
Maik Wehmeyer 2:03
Brendan, it's great to be back on the show., thanks for having me.
Brendan Le Grange 2:06
It's not uncommon for FinTech founders to come into financial services from a customer experience or a product design sort of industry. And use that experience to disrupt the old guard at least that was kind of the earlier model of FinTech, just better customer service. But your background is more technical with a statistics degree from Harvard and getting your hands dirty and real machine learning.
So before we talk about tactile and what you've built there, talk to me about your early career and where you've come from in this industry.
Maik Wehmeyer 2:36
That's right, I did start my undergrad in industrial engineering but what I enjoyed most during my undergrad was actually math and statistics. So during my graduate studies, I focused on mathematical optimization and statistics, which in 2016, meant doing a lot of machine learning.
And I fell in love.
And that is why I joined a company. So from some fellow graduate students from Harvard, called quanto, who took all of that academic knowledge and applied it to real world business challenges. And it was very interesting. And we by dad got sucked into the whole financial world of applying those models at big banks and fintechs.
And it was beautiful, because these banks and fintechs, they have a lot of financial data, very structured data. So it was a perfect playground for us as researchers to actually see what works and what doesn't work.
And we did see that a lot of these new models did actually ends up making very precise predictions on future defaults of loans on future claims of insurance policies. That was the start from the whole being of me going from engineering into in the end, fintech.
Brendan Le Grange 3:49
So what made you decide to follow that entrepreneurship path, rather than, say, taking your Harvard degree and getting into blue chip management consulting or something a little bit more traditional?
Maik Wehmeyer 3:59
Yeah, I've always been some sort of a builder. And I always like bringing people together to accomplish things. So I some of it, some of that entrepreneurial drive, since I was a kid.
So I decided not to follow my offers at McKinsey or Boston Consulting Group where I interned, because I wanted to work more technical, and I wanted to coach and I wanted to build, and yes, we were somehow consultants during the QuantCo time.
But in the end, we were building a company. And we were like trying to apply those statistical models onto problems that no one has ever applied to before and those years were incredibly entrepreneurial for me.
Brendan Le Grange 4:41
So Maik, four years in, you decided to go off and start Taktile, what was the inspiration behind that move that wanted you to take it even a step further on that entrepreneurship journey?
Maik Wehmeyer 4:54
I think it was two main insights we gained during the time I'm building a lot of machine learning models, new credit policies, loan products for big for big banks. The number one was that in order to actually operate those models, big companies, they needed infrastructure to do so.
And if you work with a big bank, and you know that, well, Brendan, there's a lot of legacy systems. And those legacy systems are not built for very modern decisioning systems, like very complex rules, systems, heuristics, machine, animals, statistical models, external data, and all of the infrastructure that we saw, actually didn't fulfil the needs that we wanted to have in order to bring the things to production that we build ourselves.
The second one was that if you help a bank to create lending IP, there's always a question around, who now understands how it works and who owns it. And the lending IP, that is something that you should own as a FinTech that you should own as a bank yourself.
And that is why we were thinking we should actually build a software that helps internal risk teams and internal credit teams to even create more lending IP faster on their own without having any external dependencies.
Brendan Le Grange 6:11
What was the reality like? And indeed at Taktile, you're across Berlin, and you're across New York now, how did that being international change the way it felt to be a young founder?
Maik Wehmeyer 6:23
The start was very tough.
We raised our first round the pre-seed round, the first week of March 2020. That was to start off COVID, it was the week where the stock market plummeted by more than 30%. And all investors told us, you're not able to raise any money for the next couple of years, because they were not predicting correctly that interest rates would go down to zero, and there would be more money in the market than ever before. But at that point in time, that week was really, really horrible. We managed to get some money.
And we also managed to get into the Y Combinator programme in San Francisco. However, we couldn't enter the country, we couldn't enter the US end of 2021, because there was a travel ban. So us building a decisioning platform for a lot of banks and fintechs, also in the US, but being located in Berlin was very tough with regards to working hours.
And one of the probably most thrilling memories I have is when we signed our first customer branch, one of the biggest micro lenders in the world. We were going live at 2am in the morning, Berlin time. You know, we were sitting there at our computers and we waiting for the first loans to hit our production API was so late, we were so tired. We were so thrilled.
But you know, you can do that for a night. You can do it for two nights, but it's not a sustainable motion to build a company. So yeah, it's quite tough to be between Berlin and New York for such a long time. But we opened the office in New York and it's going fantastic there.
Brendan Le Grange 7:49
It seems that when it comes to FinTech, and tech more broadly, Berlin is actually where the action is in Germany. So would you mind sort of shedding some light on that German FinTech scene? And guess where does it fit into the broader European ecosystem?
Maik Wehmeyer 8:03
We have offices in London and New York and in Berlin. So all of those three locations have very different advantages on building a FinTech company.
Let's start with Berlin, just very strong on engineering talent. For sure, one of the strongest pickups we have in Europe, if you look at N26, very strong engineering, a lot of people want to hire from there. So that's number one.
Then number two, a lot of things changed since Brexit. And we have seen quite grow of companies that can help you within the EU to build a FinTech company, raising bank of vivid money or mass are very big neobanks, they now have their own licensees, they can help you even with passporting, like raising through the European Union. And that, of course, is eating some lunch from the London fintechs that, you know, we're able to do that before.
Then number three, one of the main advantages in Berlin is being strongly regulated with regards to GDPR that you actually start building software with the gold standard. You know, if you go to any other country, and we do sell in Africa, in India, in Latin in the US people do consider the regulations that we have in Germany as being the gold standard for things like information security.
That is on Berlin, I think London, we're still huge fans, it's so strong on the financial ecosystem is really centred. If you're in London, like you have still all the big institutions they are you have so many big fintechs they're one of our customers like Kuder going from London, but then they have a lot of operations and building a bank in Nigeria. And you still have many of those examples of like internationally very successful FinTech still being located in London. So although there's Brexit, it's still one of the most interesting FinTech startups in the world.
But of course, New York finishing that up, that is the financial centre of the world. I've never seen a city where there's so much talent and knowledge and lending and risk. And I think that's hard to copy and hard to grow into if you're Berlin in such a short period of time.
Brendan Le Grange 10:03
You've built a software platform that allows businesses to build run and evaluate automated decision flows quicker, and in a more data driven fashion than ever before.
What does that mean in terms of nuts and bolts product?
Maik Wehmeyer 10:17
Taktile is a modern decision engine, which means that we've built a local no code platform that makes it really easy for credit and risk teams to build and run automated underwriting decisions. And doing that with a really high level of accuracy.
The amount of data that is available now has allowed lenders to really level up the accuracy of the decisions. And most lenders don't have the infrastructure they need to harness these data sources. So for me, the power of a modern decision engine like tactile is that it enables teams to leverage a wide variety of data sources, whether it's internal data, or whether it's external data, when they are building the underwriting decisions and their credit policies.
And by using Taktite, we're seeing lenders increase the amount of customers they approve, without taking any more risk. That's one huge advantage, just more customers without more risk, you do that by experimenting with new data sources, so that they can actually lend to completely new customer segments.
So if you take one of our examples, like one of our customers is a b2b credit card company in the US. And they expanding the footprint from traditional businesses, which they can easily underwrite by using bureau data to now tech startups. And they do that by using external data, which in this example, is open banking data, traditional borrows, they have very hard time to evaluate such a young company. And if they now use open banking data, they can see when that startup got new funding, and they can see how much money they're burning each month. And that in the end, gives them a very accurate insight into the financial health of those companies
Brendan Le Grange 11:59
Two months ago, or so, you were on the show talking about your first annual State of Lending report. And one of the major themes we discussed when we spoke for that was agility. So what does that mean? What does that look like, within a lender?
Maik Wehmeyer 12:14
The world is changing so fast at the moment. And all of the lenders that we work with, they want to segment risk in a much more precise manner. So having one scorecard for all of your customers is just not sufficient anymore. And it's also possible to be more accurate to go more into the details than ever before. So how fast can you actually change your scorecard? Are you dependent on your IT team? Or your depending on your engineering team? Or can you as a risk leader, as the head of lending? As a credit analyst, how fast can you actually change your scorecards without being dependent on a party like like the engineering department in your company?
Brendan Le Grange 12:56
You talk about unblocking your credit team, and I think that is a really exciting time for people in the space who have understood the value they can produce and have wanted to see it in action, but have just had to accept that there's this three month, six month 12 month delay from the point at which they design something to when they see it in operation. And that doesn't need to be there anymore. And as we said in our last session, it can't be there anymore. The world is moving too fast for for that sort of delay.
Maik Wehmeyer 13:25
That's right.
And I think on top of what you mentioning that, you know, having a good user interface by, you know, no code, low code environment, that Laos, risk experts are not software engineers, but who have deep knowledge on the risk base to actually go on their own.
On top, it's also the off the shelf data integrations that you need for address scorecards, because sometimes it's not about only adjusting a threshold or adding a new variable from the data that you're using. But it's like if I now take payroll data into account, if I now take accounting data into into account, would it actually help me to build a better predictor of future defaults or not pure accessibility to new data sources, is something which just leads automatically to more experimentation and too much faster, changing cycles of scorecards.
And it's also they don't even report it was one of the main outputs of all survey is like people are grasping for new data sources that they have not used before. And we live in a world in 2023, where there's definitely enough data out there in the market. You just got to get it to the risk teams in an easy manner. you summarise
Brendan Le Grange 14:31
It as interactive decision design. What does that term mean to you within Taktile?
Maik Wehmeyer 14:37
So at its core, interactive decision design is about ensuring that the right people in organisation can collaborate to quickly build, test and then iterate on automated decision flows. And what we briefly talked about through no code, low code, user interface, and powerful data integrations on the one hand non-technical and on the other hand technical teams can collaborate to optimise credit policies.
And this leads to you know them then undertaking highly accurate risk selection, and also leading to increased landing volumes with profitability. And the way we see this happening at our customers is actually in three ways. The first one is making complex credit decisioning. Very simple. So that means that technical and non technical colleagues can use the same tool to create an iterate on decision flows, non technical stakeholders are then empowered.
And these are the people, the risk analysts, the credit analysts that do have the deep industry and domain knowledge.
So for example, one of our customers, Save in India, they offer a product called cannot pay later, which is final pay later. But for doctors offices if people do not have health insurance, and they have brilliant and very technical engineers that build the machine learning models, but they also have the credit risk experts that are responsible for adjusting the pricing logic on top of the ML model, and having both of them on the same platform. That is one of the major advantages and why we call it into active decision design.
Even if we talk about pricing, and risk adjusted pricing price elasticities. You know, a lot of new stakeholders come into place for like within the bank of FinTech, that also, if they have the possibility, to work interactively on the decisioning platform, they can bring a lot of their ideas or even business strategies much faster into planned into production.
So if you like going to risk committee and we say we have a bigger risk appetite, now, can you actually start an A B test the next morning, and try to figure out which are the more inelastic customers in the world, or they're on your portfolio, and see how we can adjust that.
So all of that collaboration is the main thing about interactive decision design, I would say.
I would add two more. The one is leverage off the shelf data integrations, credit scores, or we talked about on banking, accounting data, payroll data, there's so much out there that you actually as a modern lending team, you want to use without, again, requiring any engineering support.
And, for example, one of our customers, Novo - it's a challenger bank in the US for SMBs. They launched the lending product with Taktile, within three months. And one of the main reasons was that we had all the data integrations already built on the platform. And you've probably also dealt with the big bureaus, and then not super, super fast, and they're not super, super modern.
But having these data integrations in place is definitely something that helps fintechs to launch fast.
And maybe the third point is that interactive decision design allows lenders to continuously improve the sophistication of the decisioning. So many of our customers, they start with something very simple, some rules, very, you know, easy system in order to make credit decisions. But over time, they want to have the possibility to combine those extra rules with predictive models.
This in the end enhances decision accuracy, and help lenders to also enhance their risk selection even further. And one great example is definitely branch offers customer by now, one of the biggest micro lenders in the world, they have very sophisticated models in place in order to predict probability of default.
Brendan Le Grange 18:25
Maik, I want to come back and just talk about one thing you mentioned there about helping fintechs launch quickly because for me, one of the defining characteristics of the tactile offering is how early in the life of a lending product you can get involved.
This is not just a product for big lenders with 1,000s of staff, you can accelerate the time to market for new products, or lenders or products going into new geographies by using credit policy templates. And these data integrations from a wide variety of countries and landing use cases.
Maik Wehmeyer 18:55
Yes, so going live with Taktile is, I would say, very smooth and can happen rapidly.
Once we sign a customer, our customer success team provides training and support for the platform. However, many of our customers prefer to just building and figuring it out on their own on the platform.
On the other hand, on top of the software, we also have customers to identify the right data providers for their use case and for that customer segment. And we do share best practices from our experiences in the industry. By now we've seen so many products and use cases that we can tell you if you want to launch a lending product in the UK. For that type of segment, I think we have a very good idea of what are the possibilities out there of data source that you can actually use in the end.
And I mentioned that before. Ultimately we believe that risk policies are the core lending IP of a FinTech or the bank and this is why we mainly provide a platform for our customers to collaborate, build and iterate on their policies. If they want some expertise. We'd love to sit down with them and share some But in the end, you as a FinTech, you should you should own all of that IP yourself. And you should iterate on your own.
Brendan Le Grange 20:06
Yeah. And you spoke there, Mike, about the ease of doing business with tactile. What does that process look like? Where should somebody who's listening to this and thinking, well, I could really do with Barton decision engine? Where should they go to start their conversation to learn a bit more about the product,
Maik Wehmeyer 20:22
www.taktile.com
Or they can connect with me on LinkedIn. But that is probably the best way to start a conversation. And from there on, we'd love to jump on a call, understand your use case, show you the product and run from there.
Brendan Le Grange 20:40
Is there a real world project that you think is a great example of what the Taktile product and service can bring to a lender?
Maik Wehmeyer 20:49
For me, still, one of the really astonishing things was, and we briefly mentioned before, was Novo - that b2b neobank in New York - where we helped them to launch the new funding product in under three months.
It's a huge company by now. And they decided to use tact and they were able to launch the novel funding product from scratch in under three months. And I think, why was that so fast. So our local platform made it super easy for the credit team and engineers to collaborate, so that they could quickly implement diverse credit policy and then bring the logic into production.
And on top, they leveraged on a data marketplace. By pulling external data from key providers, such as Experian, in order to launch a product, I think that will slowly and I'll just stick to pure speed of launching such an important product was super, super interesting. And then maybe second, one of my just favourite use cases on the platform is its branch. It's one of the most sophisticated micro lenders in the world. They are live in Nigeria and Kenya. In the end, they use very, very advanced and granular data in order to make risk positions.
So even you know, you as a customer, you can admit to that they use your cell phone data. And what they have developed in terms of sophistication in machine learning models, and they run on more than a couple of 100 1000s of underwriting decisions every day, is just astonishing. And by that they do enable financial access to consumers in countries that actually do not have a banking system that allows easy access to credit for many parts of the population.
And it's just a very, very motivating use case to see and you know, what enables a modern decisioning platform, that's probably you know, two of my personal favourites.
Brendan Le Grange 22:33
You also in there already mentioned a bit about the global data marketplace. But I wanted to talk a little bit more about it, because you were on the show, sort of two months ago now talking about the state of lending report. And one of the other key findings from that was that lenders are spending an enormous amount of their budgets for the future years, looking at ways to acquire new data sources. So you've launched this global data marketplace.
What is that and how connected it is to that finding?
Maik Wehmeyer 23:04
One of the things we saw in the state of lending report is that access to new data sources is transforming the lending industry. It was the number one request from risk leaders all over the world on what they think is transformative into the industry in this year next year, getting access to alternative data and new data sources.
So yeah, lenders are increasingly looking to leverage new insights to make better decisions. And building those data integrations can be incredibly painful and requires a lot of engineering resources. And it's one of the places that actually blocks risk teams on a day to day level, before they can really increase their risk selection. So what we now have is what we call a data marketplace. And it does remove the pain and makes it easier for lenders to integrate data sources and use those insights into their flows.
So we by now have partnered with leading traditional but also very alternative data providers like Codat from the UK for accounting data, Nova credit from New York, they building a more global credit bureau score, platform banking experience, as you know, one of the major traditional bureaus, and we now have to integrate it technically into the platform and into the product so that the people can very easily access the data and don't have to wait for weeks and months for their teams to integrate them.
Brendan Le Grange 24:29
I just want to finish by going slightly different because I saw attacked other con that you have corporate lending solutions as well, that we won't go too deep into it. But talk to me about how do those tools or philosophies differ, or do they not need to differ anymore?
Maik Wehmeyer 24:45
To be precise here, our solution is a decision engine that can be used for multiple use cases. In the end, it's up to the customer on what they want to do with it, but it is quite different the flows that you can see from the corporate lending side to be too besides, what do you see from the consumer lending side.
And the reason is that businesses are just very heterogeneous units compared to consumers. And the complexity just increases heavily. So one of the major differences that in corporate lending, b2b lending, everything is not fully automated, there's still a very long tail of businesses that need to be evaluated, that require manual review. And that is one of the things like if you look at the decision flows, if you look at the policies, people want to increase automation rates in corporate lending, they trying to get the long tail from and review from the underwriters into a more automated decision.
Plus, alternative data sources are also becoming increasingly important in b2b lending, to actually help those risk leaders to make better decisions. You know, having access to accounting data now, via code is so incredible. Having access to banking transaction data via plaid is also super incredible. So that's why I am very, very excited about b2b and 2024.
And we do see a very large influx of customers coming to us and say, we want to increase automation rates, how can we do that with you?
Brendan Le Grange 26:12
Maik, it's been a pleasure having you back on the show. It's always interesting to hear what's happening here and what's been enabled by the sort of agile systems. You mentioned earlier www.taktile.com but let's just go back there. Again, if anybody does want to have a conversation wants to reach out to you wants to read some of these use cases or look closer at their product. Where should they go to do that?
Maik Wehmeyer 26:35
We are very active on LinkedIn. Apart from that website, taktile.com please let us know if there any questions on the whole thing. I know it's an incredibly challenging time out there for fintechs. If you're just launching a new product, or you're very far in your journey of having a hundreds of 1000s of decisions every day on the platform, we would love to get in touch and discuss the case.
Overall, Brendan, it's been a pleasure again talking, throwing some ideas back and forth with you. So thanks a lot for having me on the show.
Brendan Le Grange 27:10
It's been it's been a pleasure.
And thank you all for listening.
Please do look for and follow the show on your favourite podcast platform and share the updates widely on LinkedIn where lending nerds are found in our largest concentration. Plus, send me a connection request while you're there.
This show is written and recorded by myself Brendan le Grange in Brighton, England and edited by Fina Charleson of FC Productions.
Show music is by Iam_wake, and you can find show notes and written transcripts at www.HowtoLendMoneytoStrangers.show and I'll see you again next Thursday.