AIXPERT

An agentic, multi-layer, GenAI-powered backbone to make an AI system explainable, accountable, and transparent

Consortium17 partners · 10 countries

RoleTechnical Coordinator

ProgrammeHORIZON EUROPE

Grant reference101214389

Our role

Technical Coordinator

As Technical Coordinator we steer the project's technical work and lead WP6, building the AIXPERT platform and the knowledge-graph and RAG integration that lets agents ground and explain their decisions. We also run the monitoring of agentic workflows and the project's data management.

Impact and outcomes

Targeting at least 5 AI model types integrated, retaining over 90% of their original performance
Targeting an improvement of more than 30% in user understanding of AI decisions
Targeting at least 25% better explanation quality with integrated knowledge graphs
Targeting pilots in 5 domains, with user trust up by at least 30%

The challenge

AI systems built on deep learning often work as black boxes, so users can't see how outcomes are reached. Transparency about data and training is thin, accountability for errors is unclear, and growing autonomy lacks safeguards, which undermines trust in critical settings such as healthcare and manufacturing.

Our approach

AIXPERT breaks complex AI decisions into interpretable steps carried out by GenAI-powered agents, grounded in knowledge graphs and RAG and refined through human feedback. We connect the agents to use-case knowledge sources and build the platform's explainable visualisations and monitoring. The system is validated in five pilots: healthcare, HR, manufacturing, educational robotics and music.

Website →