In the Innovation Lab, we combine in-depth industry expertise with state-of-the-art AI technology. Our hybrid AI approach combines classic machine learning models, deep learning and generative AI – for example for precise predictions, intelligent text processing or automated decision-making processes.
With agent-based analysis and response systems, we orchestrate
multi-agent frameworks that perform complex tasks independently, overcome system boundaries and can be flexibly and scalably integrated into existing environments such as SAP HANA or internal data sources.
Whether risk model, chatbot or research agent – together with our customers, our experts develop a customised strategy that sustainably improves their day-to-day digital work.
Customised AI solutions for financial service providers
We combine sound advice with state-of-the-art technologies
Our full-stack approach is based on Python, Flask and Streamlit, supplemented by powerful interfaces via REST APIs, SQL and remote functions for SAP HANA systems.
In the area of machine learning, we rely on supervised and unsupervised processes, deep learning and specific models for clustering, anomaly detection or domain-specific forecasts such as casting predictions. Generative AI and agent frameworks such as LangChain and LangGraph enable dynamic, context-based response systems – combined with KnowledgeGraphs and optional human-in-the-loop for critical decisions.
Visual models such as DocLing, OCR technologies and structured JSON output are used for document processing – ideal for analysing contracts or energy certificates.
Automated e-mail processes, report generation and chat apps round off our portfolio. This is how we create end-to-end solutions - from data integration and analysis to smart communication. Always customised, always industry-oriented.
We do not offer a ready-made product, but rather a toolbox that our customers can use to organise their daily tasks more efficiently. We work closely with our customers to identify use cases in which our range of services can create added value for any organisation.
Our customers decide on the scope of functions, the connection of the desired systems and data sources as well as the authorisation management. In this step, they work with us to define functional and non-functional requirements for their solution and outline their customised digital assistant.
Our experts work closely with our customers and their users to develop and implement a customised intelligent assistant in their IT infrastructure and ensure smooth commissioning through regular testing and further development of the solution.
A common fear in connection with AI is that decisions made by the underlying algorithms are not transparent. The transparency of decisions made is of crucial importance, especially with regard to the GDPR and the EU AI Act, when it comes to the use of AI in companies. We rely on modern, explainable ML and multi-agent processes in which decision paths, for example through thought tracing or structured agent interactions, can be reviewed transparently. This is how we ensure that our solutions are GDPR-compliant and powerful at the same time.
Conventional AI solutions require an active connection to the internet in order to function. Our customised service offering, on the other hand, is operated locally and can be easily implemented in our customers' system infrastructure without having to access the Internet to create added value. At the same time, it remains possible to allow publicly accessible information from the Internet for more general questions. In this case, however, no more information than necessary is shared with the rest of the world at any time and no data leaves the organisation.
It is true that hallucinations or false statements can occur when using artificial intelligence. This is due to the fact that certain solutions are based on a "frozen" level of knowledge that is already outdated or that ambiguous contextualisations can lead to false statements. Our range of services offers the possibility of connecting only those data sources that correspond to the truth and guarantee a clear answer. External sources such as the internet are then deliberately ignored and the artificial intelligence only uses the information that is available to it.
The automated processing of individual process steps and artificial intelligence are not intended to replace people, but to provide meaningful support in day-to-day work. By jointly analysing recurring tasks within the organisation, we identify those work steps that are particularly time-consuming and require little creativity. These process steps can be shortened by using RPA. The resulting efficiency gain allows employees to concentrate on the essential tasks. Furthermore, artificial intelligence should not be used to make decisions, but rather to provide information and solutions. The decisions still lie with our customers. AI should be seen as a support, not as a replacement for human judgement.
