Sponsored Feature Digital Banking Alternative Lending

How automation in B2B underwriting is revolutionising business lending

Taktile’s CEO Maik Taro Wehmeyer sat down with AltFi to discuss how automated decision engines are transforming the lending space 

Maik - Taktile

Maik Taro Wehmeyer/Taktile

For the first time, automating complex B2B underwriting processes is possible, and it is revolutionising the way businesses, especially SMEs, can lend to other businesses.

Taktile’s CEO Maik Taro Wehmeyer spoke to AltFi about how automated decision engines are transforming lending through the increased availability of digitised data sources, and modern technology that can, in turn, make use of this wealth of data.

Wehmeyer founded Taktile, a low-code/no-code platform that harnesses data and AI to empower credit and risk teams to make safer and smarter risk decisions alongside chief product and technology officer Dr Maximilian Eber in 2020.

It now counts leading companies including Branch, Novo, Rhino and Kueski among its customers, with a decision engine that adapts seamlessly to any use cases and with offices across New York, London and Berlin.

How is automation in B2B underwriting revolutionising business lending?

One of the reasons why automation is possible is because we have a lot more data sources available now, and especially digitised data sources. For example, we have accounting data, like Codat delivers, or open banking data from Plaid. And they have really revolutionised the accuracy and speed in which B2B lenders can underwrite. On top of those digitised data sources, the world is still very manual, and sometimes offline, especially in B2B lending.

Decision engines then enable lenders to integrate those data sources. So we need the data source to be available, and then a decision engine helps you actually make all of those data sources available to the head of lending or the head of credit for the SMB lender. What that leads to is streamlining the underwriting processes, in turn massively reducing the complexity and amount of time to do the risk assessment. 

What does this all mean for B2B lending as an industry right now?

If you look right now at the market demand, there are so many interesting new B2B lending fintechs coming up, which is always a result of possibilities in the market. Of course, there are very successful B2B lenders that have been around for a long time, but if you look at VC funding right now, a lot of funding is going into B2B fintechs. 

I find this particularly interesting, because until around four years ago, consumer fintech was very hot, but at the moment B2B in tech is super hot. And the reason is that new lenders can lend to a lot more customers at better prices, and they can scale their operations in a way that for the first time doesn't require them to hire more employees to get the job done. Previously, because everything was manual there was no way to scale your operations because you couldn't get any faster.

What challenges need to be overcome for things to be fully automated?

For me, a big challenge is how can you navigate human-in-the-loop [a model that requires human interaction] into actions with high automation rates. So building a system where you put all the applications into the bucket of making a direct decision, but then steering the difficult ones to a human when there's not enough data or not enough confidence I'd be making the right decision as a lender. 

So the integration of a manual review is one part of the puzzle and it's very important to solve well because high automation rates aren't always necessarily always a good thing. You need to increase automation rates by staying on par with your default rates. And I think the key is having a system where a human is involved in the relevant cases, but then can give feedback so that next time a person might not have to look at the case. But how you navigate that in the right way is a big challenge for B2B lenders.

Do you think the increase in automation is the reason for the surge in VC funding to B2B fintechs?

In the end, the unit economics need to make sense. So a venture capitalist wants to invest in a business which can make returns and which can scale by giving it more money. Automation in B2B lending is the only way of scaling your business. So the unit economics of a B2B lender that has high automation rates are just phenomenal. And there's a huge market need for businesses that need financing. So for me, there's direct direct correlation between higher automation rates and more VC funding.

Another macro trend that comes into play is embedded financing. Being directly integrated as a lender at the point of sale is also something which works very well. The eCommerce trend had to come first, but now you can see the second wave of fintechs that sit very closely in those marketplaces. 

There is another data source which is not publicly available, which is transaction data which you get when you own the data. So for example, if I had a small shop on Shopify, it would have a lot of cool data on how much I’ve sold over the last year, how happy my customers are, on my health as a business.

There is a lot of opportunity in making that marketplace data available to offer better financing solutions. And companies like Parafin are already doing this very successfully. It is serving all DoorDash restaurants and they can get financing directly for their business through Parafin, which is incredible.

Where do you think this is all headed for 2024?

I think in 2024 one of the biggest trends will be the use of AI to increase automation rates. My point of view is that large language models will not be able to increase the direct risk assessment because they are trained on public data on the public internet, and the public internet has not seen defaults for B2B lenders. However, these large language models are very, very powerful when it comes to creating more data, which can be used for better risk assessment, in turn increasing automation rates. 

Sometimes as an underwriter, you want to have additional information on a business and it takes a very long time to research it yourself on the internet. Now, with ChatGPT and other large language models, you can create a much more powerful data set that you can then use to make more accurate risk decisions. And rigorous decisions always give you more degrees of freedom to increase automation rates. 

We already use AI a lot for labelling and tagging for banking transactions. So cash flow based underwriting is the magic word here. It will be one of the main pillars of B2B lending in 2024. And cash flow based underwriting is only possible by making sense of these 1000s of transactions that you get from the businesses' bank accounts. And if you get 1000 transactions from a small business owner in the UK, you need to categorise the data, and you need to label it before you can do anything with it in underwriting. And that's why AI is really, really powerful.

Companies In This Article


People In This Article

Maik Taro

Maik Taro Wehmeyer

CEO And Co-Founder


More Like This