Fintech lending trends in 2026: what lenders and borrowers need to know

More and more players are going into lending. What I see in Europe this year is that more and more focus goes to BNPL, and embedded lending is also in trend. Underneath both, the thing our clients keep asking for has not changed: minimise the effort a customer spends applying for credit.

The forces reshaping fintech lending in 2026

Three things are arriving at once, and they interact.

• Clients are still trying to minimise the burden on the customer during application, and are checking additional scoring schemes to do it.
• Traceability was always an issue, and now clients are far more focused on it, especially where AI-based decision-making is involved.
• Unit economics is still the main focus, which decides how much of the decision happens before the expensive sources are called.

On the first, clients try to check additional schemes such as scoring with an understanding of the mobile phone environment: the model of the phone, the cost of the phone, and similar signals. Or other ways to minimise the burden on the customer of applying for the loan. Our machine learning work usually starts at that question rather than at the model architecture.

The European constraint is specific. Directive (EU) 2023/2225, the second Consumer Credit Directive, applies from 20 November 2026. Article 18(3) requires the assessment to rest on the consumer's income, expenses and other financial and economic circumstances, necessary and proportionate to the credit, and sets two hard limits: no special-category data under Article 9(1) of the GDPR, and social networks shall not be considered an external source. Article 18(11) adds that the assessment shall not be based exclusively on credit history, which is exactly the thin-file case.

What I also see is that our customers are trying to connect two or three alternative KYC systems and manage them in terms of efficiency, quality and price. It helps them enable the necessary flexibility for different customers, for different traffic, and to balance between them.

Unit economics is where those decisions land. Our customers try to minimise unit cost, or loan cost, and they try to do as much pre-scoring as they can before they go to the credit bureau or other expensive sources to finally confirm the loan.

AI underwriting and alternative data: how the mechanics actually work

I am completely for AI in underwriting, but it is necessary to be careful here. Traceability should be implemented in the first place. Then a very clever implementation of human in the loop, if necessary. And everything should be tested. I believe the future is full AI implementation, although for now that is not something which matches the current legislation.

That last point is not opinion. Article 22 of the GDPR gives a person the right not to be subject to a decision based solely on automated processing where it significantly affects them, and the Court of Justice held in SCHUFA (Case C-634/21) that producing the probability value is itself that decision when the lender leans on it. CCD2 Article 18(8) adds a right to human intervention and to an explanation.

From my experience, when you implement an additional innovative source for underwriting, you should understand how to raise the volume of those decision-making mechanisms during testing. In one of our projects, when we implemented an additional specific underwriting source with algorithms from the first loans, the weight of that model in the overall scoring was not more than one or two per cent. After successful testing, that volume can grow, specifically if it allows the customer's effort during application to be minimised.

Embedded lending and the disappearing origination interface

Embedded lending is in trend, and we have this kind of experience: we have built SDKs for embedded lending and APIs for embedded lending, embedded finance and embedded payments.

I believe the lending platform becomes a platform and orchestrates the ecosystem around it, with shops, with the network, with customers. Lending to the partner company is a wholesale relationship and it stops at the partner. Embedded lending means originating credit for that partner's own customers, at the moment they need it, inside the mobile application or the trade-network interface they are already using. That is a different product: your licence, your creditworthiness assessment and your duty to explain the decision, delivered through somebody else's screen. It is why embedded finance and embedded lending are growing in importance and should be taken into consideration from the first days of the project. The same shift is visible in SME banking and embedded finance.

Compliance as a delivery constraint, not an afterthought

Usually the technical side is not so difficult to implement. In terms of technical integration and implementation it is usually five to nine months depending on the project, so it is usually not a problem. The main difficult part is to go through legislation, and through the fine-tuning and implementation of the human-in-the-loop algorithm, which from one hand will be beneficial for the performance of the service and from the other hand will not bring additional risks or stop the service itself. If we take human in the loop for every decision, it makes no sense to do that. Maybe it will speed up the decision-making, but the target should be more ambitious.

From a regulatory point of view we have seen different scenarios.

1. Sometimes it stops the project even before it starts, because the customer receives a specific note from the regulator and has to solve it.
2. Sometimes we find problems with growing the traffic, because there are still steps to meet the regulations, and we run in testing or pilot mode for some time before those regulations are fulfilled and adopted within the service.
3. In one case the requirements came from the central bank. It was preliminarily agreed, and after that they asked for a meeting and added additional requirements.
4. On another project the regulatory problems came at the end. It was a difficult agreement on all the necessary documents to correspond with the regulatory requirements. It was a new market for our customer, and that is why our technical implementation was ahead of the legislation work.

That last one is the pattern to plan against, and the only defence is to start the regulatory track in the first sprint rather than the last.

Lending platforms we have delivered

Two KYC providers and full reporting, live in 5 months1.5M+ installations and 150x lending growthCar leasing, digitised end to end