About RenAI
RenAI delivers domain-specific and cross-domain foundation models, co-created with European research infrastructures and trained on curated imaging, time-series and tabular data. In a federated ecosystem aligned with the EOSC Federation, researchers can access, fine-tune and apply these models, supported by tools for data discovery, transformation, benchmarking and compliance, and by the coResearcher, an LLM-based assistant.
Grant agreement 101291378
Programme Horizon Europe, HORIZON-INFRA-2025-01-EOSC-05
Type Research and Innovation Action
Duration 1 November 2026 – 31 October 2029 (36 months)
Coordinator Athena Research Center
Why RenAI
Research Infrastructures produce vast quantities of data, but inconsistent metadata, format incompatibilities and limited quality assurance keep much of it out of reach for GenAI. Generic consumer tools lack scientific expertise and operate outside institutional governance. The EOSC Federation offers a unique opportunity to overcome fragmented and unequal access to GenAI in research.
Five interconnected pillars
Scientific Pretrained Models
Tailored, High-Impact, Domain-Ready
Models co-developed with EU RIs in atmospheric and biodiversity research, CCUS, neuroscience and cancer research. They support downstream tasks and domain-specific data synthesis, such as inferring missing values, cleaning data and reducing noise.
AI-Readiness and Transformation Infrastructure
EOSC-Aligned, Interoperable, Intelligent
Tools for discovery, quality enhancement, compliance checks and FAIR-by-design publishing, so that datasets are ready for GenAI fine-tuning and deployment within EOSC.
RenAI coResearcher
Agentic, Intuitive, Researcher-Centric
A smart, LLM-powered companion built on an agentic orchestration layer. It chains tools, models and datasets to automate workflows, from planning and discovery to cleaning, synthesis and evaluation.
Legal and Ethical Compliance Hub
Trust-by-Design for Scientific GenAI
Plug-and-play tools for GDPR and AI Act conformity, bias detection, data provenance and explainability, aligning every use case with European values and legal frameworks.
Community Upskilling
Adoption, Training, and Responsible Practice at Scale
A training and engagement programme, with workshops and RI support, that builds the skills and confidence to use GenAI responsibly.
Six objectives
O1
Develop pretrained GenAI models through strategic RI partnerships.
Six domain-specific foundation models and at least one cross-domain model, trained on curated RI data, with data synthesis capabilities.
Results: RenAI ScienceAI Models; synthetic and augmented datasets.
O2
Build an AI-readiness and transformation infrastructure.
Interoperable tools that turn raw, fragmented data into AI-usable, FAIR-compliant and legally sound resources.
Results: Data Quality, IntelliSearch and FAIR Publisher, Model Autotuner and Compliance toolkits.
O3
Empower scientific workflows with agentic AI.
A modular multi-agent architecture that chains models and toolkits into GenAI workflows, with human oversight.
Results: AgentMaster Framework; SpecialistAgents Library.
O4
Deliver the RenAI coResearcher.
An LLM-based, agent-driven assistant giving natural-language access to the RenAI ecosystem, grounded in OpenAIRE’s scholarly graph.
Results: coResearcher LLM Core; coResearcher FineTuning Corpora.
O5
Establish a federated deployment blueprint and ethical governance.
Co-designed with RIs and AI4EOSC, for secure, accountable and ethically compliant GenAI in EOSC.
Results: Deployment Blueprint; Ethical Framework and GenAI Sharing Policies.
O6
Foster community capacity and sustainable adoption of GenAI in EOSC.
Training, certification and peer mentoring tailored to EOSC nodes and RIs.
Results: RenAI Academy; MentorMatch Hub; Sustainability Report.
Approach
Phase 1
Foundations and Prototypes (M1–18)
Phase 2
Scale-Up and Integration (M19–36)
In the final 12 months, all core services are deployed in a federated set-up, with OpenAIRE as the primary deployment site.
Who is involved
Seven Research Infrastructures contribute data, policies and user communities. Environmental research: ACTRIS, LifeWatch ERIC, eLTER (via UFZ), IAGOS (via KIT) and ECCSEL. Health research: EBRAINS and ELIXIR (via CERTH and MC). Athena RC, leader of the EOSC EU Node development, coordinates. OpenAIRE provides open science corpora and scholarly communication infrastructure.
Who RenAI is for
Scientific researchers (non-AI and AI-savvy), EOSC Nodes and the Federation, Research Infrastructures, data stewards and managers, institutional repositories, EOSC and Open Science trainers, policy makers and funding agencies.