RESEARCH
Sustainable AI research for reliable systems
AI that not only sounds plausible, but also works reliably – based on company data, comprehensible and securely integrable.
Application-oriented research for explainable and trustworthy AI systems
Resource-efficient processes for large, heterogeneous and unstructured data sets
Research with direct transfer to productive business applications
Why we do research
Artificial intelligence only becomes valuable for companies when it can work reliably on their own data. Especially in business-critical processes, plausible answers are not enough: Results must be comprehensible, verifiable and traceable to reliable sources.
With autonomous AI agents, this demand is increasing further. Wer Prozesse automatisiert, ohne die zugrunde liegenden Daten korrekt, konsistent und im richtigen Kontext zu verstehen, automatisiert auch Fehler – schneller und in größerem Maßstab.
This is why moresophy is researching AI systems that not only find data, but also classify it. The focus is on the question of which data actually describes a problem, how they are connected and what they mean in the corporate context.
Our goal is AI that does not work as a black box, but as a reliable system: data-oriented, resource-efficient, explainable and confidently deployable.
Our research agenda
Reliable AI on company data
Company data is distributed across documents, databases, e-mails, specialist systems and external sources. Our research is developing methods to process this heterogeneous information in such a way that AI systems can work reliably on it.
The aim is an AI that not only generates answers, but also traces them back to existing data, sources and correlations.
Data-centric AI
Data quality
Context modeling
Traceability
Contextual data indexing
A large proportion of company knowledge is contained in unstructured information: Contracts, expert opinions, emails, damage files, test reports, guidelines or technical documents.
Our research develops processes to make this content not only searchable, but also usable in the respective context. Relevant information is recognized, condensed, linked and made available in a targeted manner for AI applications – as a basis for reliable answers, analyses and automation.
Unstructured data
Context modeling
Knowledge Graphs
Information aggregation
Relevant data rooms
Resource-efficient model orchestration
For moresophy, resource efficiency doesn’t just mean using less computing power. Above all, it is about using AI in a targeted manner: Not every question has to go directly to a large language model, and not every dataset has to be fully incorporated into the answer generation process.
Our research develops processes that narrow down relevant data spaces in advance, select suitable processing steps and only use large LLMs where they actually create added value. This creates AI systems that work more precisely, more economically and more sustainably.
Context limitation
Model orchestration
Specialized procedures
Sustainable AI operation
Hybrid AI architectures
analytical AI
generative AI
Context modeling
Controllable AI systems
Sovereign AI systems
For moresophy, digital sovereignty means more than technological independence. It is also about handling data with confidence: the ability to understand which information is relevant for a task, how it is connected and what significance it has in the specific company context.
Traditional retrieval approaches improve access to information. However, they do not automatically answer whether the information found is correct, complete or relevant to a decision in the respective context. Our research is therefore aimed at evidence-based AI systems that derive conclusions from the actual data constellation and not solely from the statistical response behavior of large language models.
This is particularly crucial for companies with distributed, local and heterogeneous data pools: sovereign AI does not require blanket data migration, but intelligent aggregation, semantic indexing and controllable processing where the data is already located.
Digital sovereignty
evidence-based reasoning
Semantic data indexing
Auditability
independent AI architectures
Research needs a foundation and exchange
Sponsored by
Since 2021, moresophy’s research and development work has been funded by the Federal Ministry for Economic Affairs and Climate Protection. The funding supports our work on AI systems that work reliably on company data: comprehensible, resource-efficient and practically applicable.
In exchange with
Reliable AI is created through dialog. That’s why moresophy works with universities, research institutions and innovation partners to develop methods that combine technological depth with concrete benefits for companies.
