AnyLog Edge Data Fabric runs AI on IoT data without cloud dependence

A new research paper from the AnyLog team introduces the AnyLog Edge Data Fabric, an agent- and edge-based platform designed to handle operational data for industrial and autonomous systems without relying on centralized cloud infrastructure. The paper, posted on arXiv, outlines how the platform presents distributed data, assets, compute resources, and services as a single logical system through components like a Distributed Metadata Layer, a Virtual Data Lake, a Unified Namespace, and the Model Context Protocol. This lets authorized users, applications, automation services, and AI agents discover and query resources without knowing where they are physically hosted. Crucially, queries and computations execute at the agents that hold the relevant data, meaning only requests and results travel across the network—drastically reducing data movement and latency.

The architecture is designed for resilience and scalability. Because operations continue locally during connectivity disruptions, the fabric supports real-time automation, Edge AI, federated learning, and resilient decision-making with no single point of failure or dependence on centralized infrastructure. Cloud systems remain useful for training, reporting, and long-term analytics, but they are kept off the critical decision path. The paper also highlights repeatable deployment from validated digital-twin configurations, allowing organizations to roll out new edge sites consistently. With 30 pages and 6 figures, the work targets IoT data management, distributed query, and decentralized data layers, positioning AnyLog as a cloud-like operating model for the edge—where data is acted upon at its source, not shipped elsewhere first.