moresophy Receives BSFZ Certification Once Again

moresophy Receives BSFZ Certification Once Again

Most AI systems today provide plausible answers. Whether these answers are also verifiable is a research question that moresophy has been working on for eight years. For this work, the company has once again received the seal of approval from the Certification Body for Research Allowances (BSFZ). Following 2021, 2022, and 2024, it is now also valid for the 2026–2027 funding period.

What the seal certifies

On behalf of the federal government, the BSFZ assesses whether a project qualifies as research and development under the Research Allowance Act. The prerequisite is the existence of scientific and technical uncertainties that cannot be resolved using existing methods and therefore require systematic research. For moresophy, this certification thus serves as independent confirmation that the company is on the right track. Three patent applications have emerged from the research conducted to date. One of these has already been granted, and two others are currently under examination.

DAPHY^®: Verifiable Answers Instead of Plausible Ones

The research focuses on DAPHY^®, moresophy’s technology for evidence-based reasoning orchestration.DAPHY^® transforms distributed enterprise data into verifiable reasoning. To do this, the technology combines data sources, analytical skills, and individual reasoning steps into a process of insight generation in which every conclusion can be traced back to concrete evidence. This creates a knowledge base for business-critical decisions that both humans and AI agents—who continue to work with the results—can rely on.

Three Research Questions

The first question concerns the transition from retrieval to structural inference. It is not enough to find relevant information. The system must also understand the structure underlying the data: its relationships and the logic according to which it is aggregated.

The second question leads from implicit reasoning to verifiable execution plans. Instead of simply producing a result,DAPHY^® models the path to that result as an execution graph that can be verified step by step.

The third question concerns the validation process itself. Instead of conducting a one-time fact-check at the end, the entire process of knowledge acquisition is continuously monitored and verified.

Each of these approaches is relevant in its own right. It is their interaction that makes the difference. That is why moresophy views DAPHY^® as a long-term research agenda for autonomous and transparent AI systems, rather than as a single function.

A continuous path of research

This work dates back to the first research grant in 2020. This was followed by the ZIM project in 2023/24 and now the current funding period. Throughout all these stages, the same conviction has held true: trust in AI is built through transparent reasoning, and larger models do little to change that.