colours

Business Automation for Swiss Banks

Automation rarely fails because of technology. It fails because no one owns the process end-to-end, because rules are buried in code instead of being maintained by the business department, and because every regulatory change requires a new release.

Five technologies, one process

BPMN and CMMN: Visually Control Processes and Cases

With BPMN and CMMN, you model and orchestrate business processes and case handling based on standards. The models are not just documentation, but the executable foundation for your workflows. This makes responsibilities, deadlines, escalations, and approvals transparent – from the very first activity to process completion.

Possible platforms: Camunda, Flowable, UiPath

Decision Management with DMN: Maintain Business Rules Yourself

With DMN, business rules are mapped in clear decision tables. Compliance and business departments can maintain rules themselves and implement changes in a controlled manner. A new threshold or verification logic is configured, not programmed. This shortens regulatory change cycles and creates a clear separation between process, decision, and technical implementation.

Robotic Process Automation: Bridge to Legacy Systems

Not every legacy system has a modern interface – and not every system will get one. RPA bridges this gap where integration is not possible in the short term or due to technical constraints. Bots handle clearly defined tasks in the front and back office. We view RPA as a pragmatic bridge, not as a replacement for sustainable system integration.

Possible platforms: UiPath, Blue Prism, Automation Anywhere

Process Mining: Measure First, Automate Second

Successful automation begins with a realistic picture of the as-is process. Process mining reveals where bottlenecks, deviations, and unnecessary repetitions occur. This helps you identify which processes are suitable for automation, where the greatest benefit lies, and how to measure improvements after implementation.

Possible platforms: Celonis, UiPath, ARIS

Agentic AI: When Rules Alone Are Not Enough

Not every work step can be fully mapped using rules: verifying documents, evaluating adverse media hits, or creating case summaries from multiple sources requires contextual understanding and human review. The process remains deterministic and auditable in BPMN. Agents work in clearly defined service tasks with specified input and output. Results are documented; regulatory decisions remain with humans. This is how hyper-automation becomes reality.

Product-neutral and transparent regarding licensing

We do not sell platforms. We choose the solution that fits your initial situation, your organization, and your regulatory requirements.

Flowable, Camunda, and Appway address different scenarios. The decisive differences are often not in the technology, but in process ownership, governance, operating model, and adaptability.

Our approach is pragmatic: where possible, we start with a Proof of Concept on an open-source version and only then decide on regulated production operations and the appropriate licensing.

You see the working process before making a long-term licensing decision. This creates transparency for your organization and a solid foundation for your business case.

Connection to Your Core Banking System

Automated processes are only as good as their integration. We connect orchestration with the system you already run (Avaloq, Finnova, Temenos, Olympic, T24) via REST/SOAP, MuleSoft, or Kafka, including IAM, e-signature, and secure document exchange.

FAQ

Both are BPMN engines with a comparable core. Flowable includes CMMN and a case model, which is often a better fit for CLM; Camunda has a larger developer community and a different licensing model. We decide based on your process profile, not on a partnership.

As a bridge, yes; as a target state, no. Where an interface is possible, we build the interface.

Your business department. That is the whole point of DMN — otherwise, we would have hard-coded it.

Anything that can be decided unambiguously belongs in DMN: there it is faster, cheaper, and fully traceable. We use agents where text needs to be read, evaluated, or summarized. Mixing the two results in a system that no one can audit anymore.

Every agent step is a process step with logged input, output, and source references. For regulatory decisions, approval remains with humans — the agent prepares, it does not decide.

More on this topic

  • Business Automation

    From 0 to AI: We turn prototypes into production reality. Learn more

  • Agentic Platforms Put to the Test

    A comparison using the example of automated address changes. Learn more

Do you have questions?

No website and no brochure can replace a personal conversation about your goals and issues. I look forward to hearing from you.

Contact