Aethon: A Reference-Based Replication Primitive for Constant-Time Instantiation of Stateful AI Agents

arXiv cs.AI / 4/15/2026

💬 OpinionDeveloper Stack & InfrastructureSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper argues that moving from stateless LLM inference to stateful, tool-using agents exposes limitations in current AI runtimes, especially around latency and memory costs from heavy instantiation.
It introduces Aethon, a reference-based replication primitive that creates stateful agent instances as compositional views over stable definitions, layered memory, and local context overlays instead of fully materializing each instance.
Aethon’s architecture and memory model use layered inheritance with copy-on-write semantics to decouple creation cost from inherited structure, enabling near-constant-time instantiation.
The authors analyze how this design affects complexity, scalability, multi-agent orchestration, and enterprise governance, positioning reference-based instantiation as a production-oriented systems abstraction.
Overall, Aethon is presented as a step toward an infrastructure pattern where agents behave like lightweight, composable execution identities that can be spawned, specialized, and governed at scale.

Abstract

The transition from stateless model inference to stateful agentic execution is reshaping the systems assumptions underlying modern AI infrastructure. While large language models have made persistent, tool-using, and collaborative agents technically viable, existing runtime architectures remain constrained by materialization-heavy instantiation models that impose significant latency and memory overhead. This paper introduces Aethon, a reference-based replication primitive for near-constant-time instantiation of stateful AI agents. Rather than reconstructing agents as fully materialized objects, Aethon represents each instance as a compositional view over stable definitions, layered memory, and local contextual overlays. By shifting instantiation from duplication to reference, Aethon decouples creation cost from inherited structure. We present the conceptual framework, system architecture, and memory model underlying Aethon, including layered inheritance and copy-on-write semantics. We analyze its implications for complexity, scalability, multi-agent orchestration, and enterprise governance. We argue that reference-based instantiation is not merely an optimization, but a more appropriate systems abstraction for production-scale agentic software. Aethon points toward a new class of AI infrastructure in which agents become lightweight, composable execution identities that can be spawned, specialized, and governed at scale.