Agentic AI and Agentic Networking for Next-Generation Computer Networks
Date, hour and room to be defined
Organisers
- Muhammad Awais Jadoon (InterDigital Europe Ltd, UK)
- Roberto Pereira (Centre Tecnològic de Telecomunicacions de Catalunya – CTTC, ES)
- Matteo Zecchin (EURECOM, FR)
- Mateus Pontes Mota (CEA-Leti, FR)
- Dariush Salami (Nokia Bell Labs, FI)
- Rasoul Behravesh (Net Reply, UK)
- Farhad Rezazadeh (Hostelworld Group, ES)
Motivation and Background
The transition toward 6G marks a fundamental shift from connectivity-oriented networks to AI-native, autonomous, and goal-driven systems that tightly integrate communications, computing, and sensing. While 5G introduced AI-assisted functions, intelligence largely remains reactive, and constrained within predefined procedures, and current intelligence remains largely “plugin”: it can monitor and predict, but it typically cannot execute multi-step, autonomous closed loops across network domains and time scales.
Agentic AI introduces a disruptive paradigm for computer networks. Instead of treating AI as an external optimization tool, Agentic AI embeds autonomous, reasoning-capable AI agents directly into the network fabric. Such agents can perceive multimodal context, interpret intent, plan and execute actions, and coordinate with other agents across the network. This enables a transition from static, function-specific control loops to distributed, self-adaptive intelligence. Key areas of applicability are 6G-related services such as sensing, computing, and digital twins for vertical applications and network management.
In the context of computer networks, Agentic AI raises new architectural, protocol, and governance challenges. These include agent discovery and communication, exposure of network capabilities as agent tools, real-time coordination under strict latency constraints, safety and accountability of autonomous decisions, sustainability, and alignment with ongoing standardization efforts across bodies such as 3GPP, ITU, IETF, ETSI, AI-RAN, and O-RAN. Additional challenges include biased decision-making and unstable or unfair behaviour, which can be difficult to detect, correct, and govern.
This workshop aims to provide a focused forum on Agentic AI for next-generation computer networks, addressing both agents-for-network and network-for-agents perspectives. The scope includes architectures for agentic networking, agentic use cases, AI-native control and management, agent communication and orchestration, sustainability, and implications for standardization. By bringing together researchers, industry practitioners, and standards experts, the workshop seeks to identify research gaps, align emerging efforts, and outline a practical roadmap toward realizing agentic, AI-native networks.
Structure
All invited speakers and TPC members had been firmly confirmed before the workshop proposal was submitted. The ITU-T FG AINN management team agreed to the keynote and will confirm the individual and keynote title once the workshop is accepted. The format is Keynote talk, accepted papers presentations and invited talks. The breakdown and details are as follow:
Time 0:00 – 0:05
Welcome by workshop moderator
0:05 – 0:30
Keynote, 20min + 5min Q&A
(confirmed) Representative of the ITU-T FG AINN Management Team, Title: TBC
0:30 – 1:30
Accepted Paper 1, 12min + 3min Q&A
Accepted Paper 2, 12min + 3min Q&A
Accepted Paper 3, 12min + 3min Q&A
1:30 – 1:45
Break
1:45 – 3:00 (Invited Talks)
Invited Talk 1, 15min
Dr Diego R. Lopez (Telefónica Innovación Digital)
Data as the Foundation for Growing Value
Invited Talk 2, 15min
Dr Sebastian Robitzsch (InterDigital Europe Ltd)
Challenges and Opportunities on Agentic Networking for AI Agents in 6G
Invited talk 3, 15 min
Merim Dzaferagic (Trinity College Dublin, Ireland)
From Architecture to Deployment: Supporting Agentic AI Use Cases in Distributed 6G Networks
Invited Talk 4, 15min
Dr Nicola Piovesan (Huawei Technologies, Paris)
Toward AI RAN: Agents for Autonomous Networks
Invited Talk 5, 15min
Dr Davit Harutyunyan (Aigentifyable, Bosch)
Agentic AI Networks: From Reactive Automation to Autonomous Intelligence
