Openfund

CORDIS · CoAgent

Cognitive Computing Continuum for Large-Scale Distributed GenAI Agents

Код
CoAgent
Програма
HORIZON
Процедура
HORIZON-CL4-2025-04-DATA-02
Статус
SIGNED
Период
01.10.2026 – 30.09.2029
Обща стойност (целият консорциум)7 944 936 €
Принос на ЕС (целият консорциум)7 944 936 €
Принос на ЕС за българските участници334 250 €

Официални данни · CORDIS · синхронизирано 17.10.2026 10:16 · официален запис

Участници

ОрганизацииРоляДържаваБезвъзмездна помощ
UNIVERSITAET INNSBRUCKкоординаторAT756 875 €
SINTEF ASучастникNO902 125 €
TELENOR ASAучастникNO632 356 €
PHILIPS MEDICAL SYSTEMS NEDERLAND BVучастникNL607 625 €
ONLIM GMBHучастникAT556 250 €
CONSORZIO INTERUNIVERSITARIO NAZIONALE PER L'INFORMATICAучастникIT534 250 €
ALMAVIVA THE ITALIAN INNOVATION COMPANY SPAучастникIT520 645 €
CLOUDFERRO SAучастникPL425 500 €
BEAWRE DIGITAL SLучастникES407 500 €
ETRA INVESTIGACION Y DESARROLLO SAучастникES392 375 €
INDRA SISTEMAS SAучастникES382 500 €
AUCKLAND UNIVERSITY OF TECHNOLOGYучастникNZ380 000 €
NEVROMORFIKA EOODучастникBG334 250 €
NEXTWORKS SRLучастникIT327 375 €
University Industry Foundation Yonsei UniversityучастникKR306 250 €
SEOUL NATIONAL UNIVERSITY R&DB FOUNDATIONучастникKR272 500 €
COMUNE DI MESSINAучастникIT206 560 €
BEAWRE POLSKA SP Z.O.Oтрета странаPL0 €
ALMAWAVE SPAтрета странаIT0 €
PHILIPS ELECTRONICS NEDERLAND BVтрета странаNL0 €

Описание

AI agents show great potential in solving real-world problems, but are hard to scale. Some of the reasons are their heavy resource consumption, brittle collaboration under distributed infrastructures, and the inherent AI uncertainty of AI lacking human trust.CoAgent researches a new agent continuum that enables GenAI agents to operate seamlessly across trustworthy cloud-edge infrastructures via three groundbreaking research pillars: (1) Agent Operation for intent-based orchestration of models and agents cross edge, cloud and HPC, exploiting small and unconventional devices with energy efficiency; (2) Agent Connectivity for autonomous agent collaboration and distributed decision making, supported by on-demand network and tool use; and (3) Agent Trustworthiness for secure execution, continuous observability, risk governance, and human-agent alignment.The results are 12 enablers that help agentic AI application developers and infrastructure providers to build a cognitive continuum, validated by four at-scale pilots for mobility management, medical device maintenance, drone mission control, and telecom operation, extended to biomedical research and smart cities.Impact-wise, CoAgent aims to reduce Europe’s dependency on US cloud providers, currently operating over two-thirds of its GenAI workloads, to less than one-third by enabling GenAI agents to run across the cloud-edge continuum and promoting EU cloud and models. CoAgent gathers a consortium of 17 partners (plus 4 affiliated entities) from 9 countries. Five world-leading industry groups drive the large-scale uptake of agentic AI and CoAgent results, with five SMEs as technology suppliers, supported by five universities, research institutes, and one public sector organization. The consortium covers a complete value chain for EU-made continuum, provides access to EU GenAI models and agentic platforms, and represents strong international collaboration with the Republic of Korea and New Zealand.

Официални данни · CORDIS · синхронизирано 17.10.2026 10:16 · официален запис · Лиценз: © European Union, CC BY 4.0 (Commission reuse policy).