Agentic Federated Learning: The Future of Distributed Training Orchestration

Rafael O. Jarczewski; Gabriel U. Talasso; Leandro Villas; Allan M. de Souza

cs.MAApr 6, 2026

Agentic Federated Learning: The Future of Distributed Training Orchestration

Authors: Rafael O. Jarczewski, Gabriel U. Talasso, Leandro Villas, Allan M. de Souza

Organizations: Institute of Computing, University of Campinas and Hub of Artificial Intelligence and Cognitive Architectures - H.IAAC, Campinas, Brazil

Abstract

Although Federated Learning (FL) promises privacy and distributed collaboration, its effectiveness in real-world scenarios is often hampered by the stochastic heterogeneity of clients and unpredictable system dynamics. Existing static optimization approaches fail to adapt to these fluctuations, resulting in resource underutilization and systemic bias. In this work, we propose a paradigm shift towards Agentic-FL, a framework where Language Model-based Agents (LMagents) assume autonomous orchestration roles. Unlike rigid protocols, we demonstrate how server-side agents can mitigate selection bias through contextual reasoning, while client-side agents act as local guardians, dynamically managing privacy budgets and adapting model complexity to hardware constraints. More than just resolving technical inefficiencies, this integration signals the evolution of FL towards decentralized ecosystems, where collaboration is negotiated autonomously, paving the way for future markets of incentive-based models and algorithmic justice. We discuss the reliability (hallucinations) and security challenges of this approach, outlining a roadmap for resilient multi-agent systems in federated environments.

Figures & tables

Figure 1: AgenticFL Paradigm: shows how an agent or a multi-agent system can be introduced into the FL pipeline.

Cifar 10
Models	Prompt	PoC			Random			Oort
	Prompt	K avg	Acc. (%)	ST (s)	K avg	Acc. (%)	ST (s)	K avg	Acc. (%)	ST (s)
Qwen3	DO	7 $\pm$ 5	38 $\pm$ 13	3,79	6 $\pm$ 5	39 $\pm$ 11	4,71	12 $\pm$ 6	39 $\pm$ 14	4,45
	Few-Shot	6 $\pm$ 4	38 $\pm$ 13	1,43	5 $\pm$ 4	38 $\pm$ 15	2,28	9 $\pm$ 5	38 $\pm$ 15	2,96
	CoT	7 $\pm$ 5	39 $\pm$ 13	1,79	5 $\pm$ 4	37 $\pm$ 14	1,65	9 $\pm$ 4	39 $\pm$ 13	2,53
Llama3.1	DO	10 $\pm$ 0	37 $\pm$ 13	0,18	10 $\pm$ 0	38 $\pm$ 10	0,38	9 $\pm$ 2	38 $\pm$ 13	0,19

Table 1: Impact of LLM and Prompt on Selection and Performance

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 3: Convergence of PoC : The interval highlights the dynamic adaptation of the agent’s strategy.

Figure 4: Experiment comparing raw LLM to random selection and ToolAgent.

N° Clients	Approachs	Completion Tokens	Prompt Tokens	Total Cost ($)	Total Tokens	Accuracy
5	LLM	178 $\pm$ 52	1336 $\pm$ 15	0.000308 $\pm$ 0.000031	1515 $\pm$ 54	0.766264 $\pm$ 0.105471
5	Tool Agent	164 $\pm$ 122	6193 $\pm$ 1963	0.000804 $\pm$ 0.000241	6358 $\pm$ 2072	0.792612 $\pm$ 0.101721
10	LLM	218 $\pm$ 56	2259 $\pm$ 28	0.000470 $\pm$ 0.000035	2478 $\pm$ 64	0.693686 $\pm$ 0.066431
10	Tool Agent	289 $\pm$ 218	8367 $\pm$ 3103	0.001149 $\pm$ 0.000431	8657 $\pm$ 3297	0.785554 $\pm$ 0.085655
25	LLM	254 $\pm$ 67	5035 $\pm$ 73	0.000908 $\pm$ 0.000041	5289 $\pm$ 97	0.637390 $\pm$ 0.090126
25	Tool Agent	248 $\pm$ 244	8056 $\pm$ 3275	0.001080 $\pm$ 0.000484	8305 $\pm$ 3493	0.882115 $\pm$ 0.146230

Appendix

Table 2: The number of tokens used according to the increase in the number of clients by LLM .

Federated Learning

Federated learning trains shared models across data holders without requiring them to pool their raw data.

See all 921 papers

Agentic Federated Learning: Rule-Based Client and Server Agents for Adaptive Training
Sep 28, 2026Deepthy K. Bhaskar, VP Binu, B MinimolFederated Learning Aggregation Client Selection in Federated Learning
Department of Computer Engineering, Model Engineering College, APJ Abdul Kalam Technological University, Thiruvananthapuram 695016, Kerala, India. · Department of Biomedical Engineering, Model Engineering College, APJ Abdul Kalam Technological University, Thiruvananthapuram 695016, Kerala, India.
FL-MAESTRO: Multi-Agent LLM Orchestration for Resource-Constrained Federated Learning
Aug 20, 2026Jiajun Wu, Zirui Wang, Jiayu Zhou +2Federated Learning Federated Learning Aggregation
Department of Electrical and Software Engineering, University of Calgary, Calgary, Canada · School of Information, University of Michigan, Ann Arbor, MI, USA
Cooperative Multi-Agent Reinforcement Learning for Adaptive Aggregation in Semi-Supervised Federated Learning with non-IID Data
Aug 26, 2026Rene Glitza, Luca Becker, Rainer MartinFederated Learning Federated Learning Aggregation
Institute of Communication Acoustics, Ruhr-Universität Bochum, Bochum, Germany

Sep 28, 2026J/K move · Enter open · S save

Agentic Federated Learning: Rule-Based Client and Server Agents for Adaptive Training

Deepthy K. Bhaskar, VP Binu, B Minimol

Department of Computer Engineering, Model Engineering College, APJ Abdul Kalam Technological University, Thiruvananthapuram 695016, Kerala, India. · Department of Biomedical Engineering, Model Engineering College, APJ Abdul Kalam Technological University, Thiruvananthapuram 695016, Kerala, India.

Federated Learning Aggregation Client Selection in Federated Learning Non-IID Federated Learning Federated Learning

Open paper AskPDF arXiv

Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it suitable for privacy-sensitive applications such as healthcare, finance, and edge intelligence. However, conventional FL approaches rely on static client participation and fixed aggregation strategies, which limits their effectiveness under non-IID data distributions, heterogeneous client behavior, and noisy or unreliable updates. To overcome these issuess, this paper proposes an Agentic Federated Learning (AFL) framework that integrates lightweight rule-based autonomous agents at both client and server levels. The proposed framework introduces a Client-Side Agent (CSA) that dynamically adapts local training parameters, controls participa- tion, and evaluates update reliability, while a Server-Side Orchestrator Agent (SSOA) performs quality-aware client selection and adaptive aggregation. Unlike traditional FL methods, AFL enables context-aware decision-making during the training process, improving adaptability and robustness in dynamic distributed environments. Extensive experiments conducted on the CIFAR-10 dataset under IID, non-IID, and noisy-client settings demonstrate that AFL consistently outperforms standard base- lines including FedAvg and FedProx. Experimental results show improvements in classification accuracy, convergence speed, robustness against corrupted updates, and communication efficiency. Ablation studies further confirm the complementary contributions of CSA and SSOA, while statistical analysis validates the significance of the observed gains. The proposed AFL framework demonstrates that incorporating autonomous agentic reasoning into federated learning provides an effective and practical solution for intelligent, adaptive, and robust distributed learning systems.

Abstract

Federated Learning

Explore similar work

Agentic Federated Learning: Rule-Based Client and Server Agents for Adaptive Training

FL-MAESTRO: Multi-Agent LLM Orchestration for Resource-Constrained Federated Learning

Cooperative Multi-Agent Reinforcement Learning for Adaptive Aggregation in Semi-Supervised Federated Learning with non-IID Data