Akkuro Lending streamlines Swiss digital SME lending with Azure Red Hat OpenShift

In modern lending, the bottleneck isn’t demand. It’s decision readiness. Lending decisions depend on documents that arrive in inconsistent formats: PDFs, spreadsheets, scans, which must be interpreted and validated, often manually. When that information isn’t immediately usable, decisions stall and outcomes become inconsistent.

Akkuro, by Topicus builds financial software for complex lending operations across Europe. The Akkuro Lending platform is designed to make document intelligence a native part of the workflow so that there is no wait for manual review to reach a decision.

In Switzerland, strict regulation adds constraints: financial data must stay within national borders, every step must be auditable, and institutions can’t send sensitive client data to external systems for processing.

For lenders, the tradeoff has been familiar: move faster and risk noncompliance or stay compliant and accept that decisions take weeks.

Getting document intelligence right, within required data boundaries, at production scale and on documents, is what separates viable digital lending from systems that stall. Traditional optical character recognition (OCR) wasn’t reliable for the task. 

Turning documents into decisions, faster

Microsoft Azure Red Hat OpenShift is the platform foundation for Akkuro, providing a consistent, highly secure Kubernetes environment where lending workflows, data handling, and document intelligence run together at production scale, eliminating handoffs that slow approvals.

When an application is submitted, financial data is interpreted and structured automatically. By the time a case is opened, the data is usable. But building that capability wasn’t straightforward.

Kamuno, a Swiss digital lender backed by Urner Kantonalbank, was the first to run the full Akkuro Lending platform in Switzerland. It launched with no historical data, so the system had to learn and improve in production.

“We had to design the system to learn as it runs. And because AI is constantly changing, the system you design today is probably out of date next month. How do you build a future-proof production system?” asks Marco van de Haar, Head of Engineering at Akkuro, by Topicus.

This small team moved from concept to production in roughly four months, building for continuous evolution rather than static optimization.

Built for sovereignty, without slowing down

Azure Red Hat OpenShift provides a standardized, enterprise-grade deployment model, making consistent, compliant deployment possible across regulated markets. Using this solution, Akkuro replicated its EU environment in the Switzerland North region in weeks, without rebuilding. 

“Azure Red Hat OpenShift provided us with a foundation to move with speed, while being robust enough to run in a serious banking environment. Trust, reliability, and scalability is our license to operate," says Jamie Burink, Chief Growth Officer at Akkuro, by Topicus.

Microsoft Foundry runs on that same OpenShift platform, using Azure OpenAI in Foundry Models to interpret unstructured financial documents, such as PDFs and scanned forms. The solution converts these into structured, usable data within the lending workflow, with processing kept in-country to support Swiss data sovereignty requirements.

With in-country processing, loan officers open cases with financial data already extracted and structured, while keeping document processing within Switzerland. 

“We had a hard requirement that all data be stored and processed in-country,” adds van de Haar. “We needed the platform to run the same way, within those boundaries.”

A platform designed for repeatability and scale

Azure Red Hat OpenShift serves as the orchestration layer for Akkuro’s core services. This makes the deployment model repeatable across regions, including Switzerland, without redesigning the platform. Alongside the OpenShift‑hosted core, Akkuro also runs supporting services on Azure Kubernetes Service (AKS) to handle noncore workloads without overloading the primary platform.

“Some of our applications have strict technical requirements. That’s why we use Azure SQL Managed Instance and Azure Virtual Machines to run those components, so we keep everything connected within the same governed Azure environment,” explains van de Haar.

As a fully managed service jointly operated by Microsoft and Red Hat, it reduces operational complexity while maintaining the flexibility of upstream OpenShift, making it possible for Akkuro to scale consistently while maintaining compliance and operational control.

“The biggest part of our lending solution runs on OpenShift. We could just take it up, copy it, put it there, and it just runs. OpenShift supports that, and because it runs in Azure, you have regions for Switzerland, so we could deploy there, complying with all the data regulations,” says van de Haar.

Together, these services help Akkuro meet regulatory, operational, and performance requirements:

Azure Red Hat OpenShift: Core platform for Akkuro Lending’s business-critical lending services, providing a standardized and repeatable deployment model across regions.
Foundry: Document intelligence on OpenShift to extract and structure financial data inside the workflow, with processing kept in-country for Swiss data sovereignty.
Azure SQL Managed Instance: SQL data layer for structured lending and application data.
Azure Virtual Machines (Windows Server): Compute for workloads with specific technical requirements that are not yet containerized, supporting a phased modernization approach.
AKS: Kubernetes runtime for supporting and ancillary services adjacent to the OpenShift-hosted core platform.

From hours of effort to automated processing

In traditional lending workflows, processing financial statements can take up to four hours per application, which is repeated across portfolios and ongoing reviews.

With Akkuro’s approach, 70% to 90% of documents are processed without manual intervention. Teams focus on exceptions requiring human judgment rather than routine extraction.

In one production evaluation, the system achieved a median accuracy of 83%. For inconsistent, unstructured financial documents with no standardized format, that level of accuracy represents meaningful automation of work that previously required trained analysts.

A blind evaluation capability helps the team improve by testing prompt changes against real production documents without engineers accessing underlying customer data. What once took weeks now takes days.

When the data is ready, decisions follow

The impact of intelligent document processing shows up in Akkuro in three practical ways: faster decisions, higher confidence in the data lenders start with, and real operational relief. Document-heavy work no longer dictates the pace of the process. Information becomes structured and usable earlier in the workflow, so teams spend less time on manual sorting and extraction and more time on exceptions that require judgment.

Kamuno today uses Akkuro Lending to run a fully digital small and medium-sized enterprise (SME) lending operation, meeting Swiss banking regulations and in-country data requirements. Operating with an intentionally lean team and no legacy infrastructure, Kamuno set a nonnegotiable requirement from the start: deliver complex financing through a highly automated, low-friction operation. Automated credit decisioning was built into the business from day one.

“It’s not just about speed. It’s about giving businesses an answer faster and knowing it’s accurate. Microsoft Foundry helped deliver exactly that for Akkuro, which has created significant value for Kamuno’s SME clients,” says Regis Lehmann, CEO at Kamuno.

Speed and regulation are often treated as tradeoffs, but the Akkuro, by Topicus approach shows the real constraint is where the work happens. When document interpretation runs inside the workflow, within required boundaries, decisions move faster with the auditability regulators expect.

Looking ahead, Burink sees intelligent document processing (IDP) as more than a document capability. 

“Today, IDP helps banks turn document-heavy work into decision-ready workflow input,” says Burink. “Over time, that same foundation supports the shift from documents to proof and from static workflows to more intelligent, agentic banking operations.”

Discover more about Akkuro, by Topicus on LinkedIn.