Accelerating Trustworthy AI in Drug Discovery with ELSA
Earlier this year, Kiin Bio were honoured to be selected as one of seven winners of the European Lighthouse on Secure and Safe AI (ELSA) 2nd Industry Call. This recognition marked an exciting milestone for Kiin Bio and an important early endorsement of the platform we are building.
Through this collaboration, we advanced our AI Platform for Experimental Design, achieving major scientific, technical, and community milestones that bring us closer to our mission: accelerating therapeutic discovery through intelligent, explainable, and human-centred AI.
About ELSA
The European Lighthouse on Secure and Safe AI (ELSA) is a European network of excellence dedicated to advancing the foundations of trustworthy, transparent, and human-centred AI.
Funded by the European Union and coordinated by leading research institutions, ELSA connects academia, startups, and industry to drive responsible innovation in areas such as healthcare, robotics, cybersecurity, multimedia, and document intelligence.
As part of its mission to translate cutting-edge research into real-world impact, ELSA launched its Industry Calls to support startups developing ethical and secure AI solutions. Each call provides up to €60,000 in funding, mentorship, and access to a thriving innovation ecosystem.
Increasingly pervasive deployment of AI systems, often building upon machine learning, have highlighted the urgency of enforcing the principles of Trustworthy AI to make these systems work for the good of the people and society. Achieving this goal requires societal and policy actions, but also research in technologies and social principles.
The ELSA Industry Calls enable the top researchers in the ELSA network to contribute their expertise to the development of applications and thus to advance the projects of the Industry Call winners significantly. In turn, they gain important insights into the needs of end users and of the industry, says ELSA-Coordinator Professor Mario Fritz.
The Challenge
Modern drug discovery remains slow, costly, and complex due to the vast biological and chemical spaces that scientists must explore. While AI offers enormous potential, many systems still lack the transparency, interpretability, and scalability required for real scientific use.
At Kiin Bio, we set out to address this challenge by building an AI platform that researchers can trust, understand, and collaborate with, helping them design and execute experiments faster and more accurately.
Our Approach
With ELSA's support, we delivered a multi-workstream project designed to strengthen the scientific and technical foundations of our Virtual Scientist platform.
Over the six-month collaboration, we achieved several key milestones:
Migrated our infrastructure to support long-running Virtual Scientist workflows
Enhanced reasoning accuracy through iterative schema and feedback loop refinement
Expanded our toolset, adding more than 50 new tools and extending support for NVIDIA BioNeMo and nf-core pipelines
Developed VSE-Bench, one of the first benchmarks for evaluating AI reasoning in experimental design
Supported the growth of our community, deepening collaborations and securing additional funding
The use of AI in health poses significant demands on AI's safety and security. These range from robustness over privacy to explainability to general responsible use of AI with human oversight. In this complex and demanding field, Kiin Bio is pioneering work in drug design that leverages the latest agent-based AI technology. The interaction with the Kiin Bio team has been truly inspiring. Not only have they made substantial, impressive progress within the ELSA Industry Call. They also meaningfully contributed to ELSA by sharing their experience and knowledge at the ELSA General Assembly 2025 and actively networking with ELSA's academic and industrial members.
Mario Fritz,
ELSA-Coordinator
The Impact
Through the ELSA project, we advanced our platform into a scalable, community-driven system that combines data-driven reasoning, human feedback, and modular scientific tooling.
Within just six months, we:
Improved experimental design accuracy beyond state-of-the-art models
Reduced workflow latency and increased reliability across complex, multi-tool experiments
Established transparent and auditable AI workflows to promote trust and reproducibility
Empowered researchers to accelerate discovery using explainable, collaborative AI
2025 ELSA General Assembly Delegates. Photo © Sandra Engel/CISPA
Our results were showcased at the 2025 ELSA General Assembly in Brussels, where we presented "Disrupting Drug Discovery with Virtual Scientists". The event brought together European AI leaders and policymakers and highlighted how explainable AI can drive responsible innovation in life sciences.
Kiin Bio presenting "Disrupting Drug Discovery with Virtual Scientists" at the ELSA General Assembly Meeting. Photo © Sandra Engel/CISPA
Since its founding, ELSA – European Lighthouse on Secure and Safe AI has focused on leveraging its broad industrial participation by developing two pillars: One is the ambitious use cases, which cover a broad spectrum of sectors where the real-life impact of safe and secure AI is expected. The second is ELSA's Industry Calls. On top of the exchange between academia and industry within the Use Cases, they broaden the range and scope of information exchange among academic and industrial ELSA network members and various sectors. The exchange with young entrepreneurs working on innovative AI applications is of great value to ELSA. AI remains a technology with substantial risks, and it is on responsible players to innovate in line with European and societal values and to decide where and how to use it to leverage its potential for societal good.
What's Next
Building on this momentum, we are continuing to:
Expand integrations into metabolomics and imaging analysis
Release components of the VSE-Bench dataset for open benchmarking
Deepen collaboration with academic and industry partners
Scale access to the platform for research institutions and enterprises in Europe and beyond
Looking Ahead
The ELSA Industry Call has helped accelerate our journey towards creating a new standard for ethical and explainable AI in science.
By combining human expertise with transparent machine reasoning, we are helping shape a future where AI becomes a trusted collaborator in discovery, advancing both innovation and integrity across biomedical research.