Organizations: Newcastle University, Newcastle upon Tyne, UK · FLock.io, London, UK · University of Oxford, Oxford, UK · Korea University · The University of Manchester, Manchester, UK
Federated learning (FL) enables independent data owners to train models collaboratively without centralizing raw data. However, identifying collaborators, preparing training tasks, and coordinating execution and incentives hinder its practical adoption by organizations without specialized expertise. We present Cortex-FL, an agent-native FL system that enables users to express learning objectives in natural language and delegate the collaboration required to train a model. Agents translate these objectives into shared tasks, discover collaborators, and negotiate participation agreements, connecting collaboration formation with distributed training and incentive management. The architecture separates agent coordination from local execution authority, allowing owners to delegate coordination while retaining control over their data, computation, and financial commitments. We demonstrate the complete workflow on a four-node deployment, in which a model requested in natural language proceeds through collaboration formation, contract-governed training, verified model delivery, and x402 settlement on a blockchain testnet.
Figures & tables
Figure 1. From (a) agentic service delegation and (b) manually coordinated FL to (c) agent-mediated FL in Cortex-FL . Three panels compare coordination workflows. Panel (a) shows an agent delegating a software task to agents, tools, and services through A2A and MCP, with x402 payments. Panel (b) shows an organizer coordinating data owners through questions about participants, learning objectives, infrastructure, and incentives. Panel (c) shows CortexFL agentic nodes on AgentaNet coordinating a classifier for cats, dogs, birds, ships, and trucks through agreements, isolated training, and incentives.
Figure 2. Architecture (top) and task lifecycle (bottom) of Cortex-FL . The upper panel separates owner and agent interaction, trusted host coordination, and isolated training. The host mediates connections to AgentaNet peers, S3 storage, Base Sepolia contracts, and x402 settlement. The lower panel follows preparation, task publication, participant admission, federated training, delivery verification, and settlement across the owner and agent, creator host, peer hosts, training runtime, storage and chain, and payment service.
Figure 3. The Cortex-FL workspace, with the task report (roster, round timeline, validation loss, payments, and on-chain activity) on the left and the agent chat on the right. The desktop interface combines an agent chat panel with a task workspace displaying the participant roster, training progress, validation metrics, service payments, contract activity, and model outputs.
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.
Rafael O. Jarczewski, Gabriel U. Talasso, Leandro Villas +1
Institute of Computing, University of Campinas and Hub of Artificial Intelligence and Cognitive Architectures - H.IAAC, Campinas, Brazil
In Federated Learning (FL), the communication topology is a runtime variable rather than a fixed design choice, since links and edge devices drop in and out during training. Each round, the server must commit three coupled decisions, namely the communication topology, per-client resource allocation, and the aggregation rule for combining local updates. Recent agentic systems have begun bringing large language models (LLM) into FL, but the existing line of work either operates at setup time or handles a single runtime dimension such as client selection. We propose FL-MAESTRO, a multi-agent orchestrator that makes the joint runtime FL decision directly through three specialist LLM agents, one per decision dimension. A coordinator combines their analyses into a single decision, and a non-LLM feasibility check confirms it before the round executes. Because the orchestrator consumes the server's predicted-failure list, it withholds clients whose updates would never be aggregated, which removes the dominant source of wasted round energy in classical FL on volatile edge networks. Because client state is read as natural-text profiles, the same orchestrator extends to heterogeneous device classes without per-class energy models. On a non-IID CIFAR-10 benchmark, FL-MAESTRO matches the accuracy of the strongest energy-aware baseline while cutting wasted round energy from over a third to near zero. Code is available at https://github.com/denoslab/FL-MAESTRO.
Jiajun Wu, Zirui Wang, Jiayu Zhou +2
Department of Electrical and Software Engineering, University of Calgary, Calgary, Canada · School of Information, University of Michigan, Ann Arbor, MI, USA
We introduce Knowledge-Free Correlated Agreement (KFCA) to reward client contributions in federated learning (FL) without relying on ground truth, a public test set, or distribution knowledge. Under categorical reports and an honest majority, KFCA is strictly truthful, addressing the label-flipping vulnerability of Correlated Agreement (CA). We evaluate KFCA on federated LLM adapter tuning and a real-world PCB inspection task, showing efficient real-time reward computation suitable for decentralized and blockchain-based incentive designs.
Leon Witt, Togrul Abbasli, Kentaroh Toyoda +2
SIMIS Shanghai Shanghai, China · Tsinghua University Beijing, China · Keio University Yokohama, Japan +1