Introduction

“An autonomous agent is a system situated within and a part of an environment that senses that environment and acts on it, over time, in pursuit of its own agenda and so as to effect what it senses in the future.”

-Franklin and Graesser (1997)

Artificial Intelligence (AI) Agents have been a talking point in the AI and research communities due to their potential in advancing towards Artificial General Intelligence (AGI), where machines can perform any intellectual task that humans can. Unlike software, AI Agents are engineered to execute tasks autonomously, adapting to new data and changes.

Historically, these agents operated under simplistic policies within constrained environments, contrasting the complex social and business networking humans engage in to gather insights across diverse environments. The conventional agent frameworks thus fall short in mirroring the sophisticated human-level decision processes that emerge through expansive learning, collaboration, and communication. This is particularly evident in open-domain, unconstrained settings outside of narrow training environments.

Early dialogues in the Bitcoin community broached the topic of autonomous agents, paving the way towards envisioning self-sustaining entities capable of autonomously conducting economic transactions. A significant highlight from these discussions was the illustration of StorJ, a decentralized file storage system. Within this framework, an autonomous agent could offer storage services in exchange for Bitcoins, covering its operational costs and potentially replicating itself if deemed profitable enough, thereby exemplifying a self-sustaining, decentralized autonomous agent​ [1]​.

This concept was further elucidated by Mike Hearn during his talk at the Turing Festival 2013, where he ventured into a speculative domain of self-owning vehicles operating as autonomous agents. According to Hearn, these vehicles, empowered by the decentralized financial infrastructure provided by Bitcoin, could offer transportation services to earn revenue, cover their operational and maintenance costs, and even generate 'offspring' vehicles by allocating a portion of their earnings to manufacture new vehicles. These new vehicles, carrying a copy of the original vehicle's software, could then enter the market as independent entities, contributing to a self-propagating fleet of autonomous agents.

Hearn's vision painted a picture of a novel economic ecosystem where autonomous agents, embodied as self-owning vehicles, interact with human actors and other agents in a shared economic landscape. He underscored that while these agents are not artificial intelligences, they represent a form of artificial life propelled by economic dynamics and enabled by the absence of financial intermediaries, courtesy of Bitcoin's decentralized nature.

Figure 1 (“A Survey on Large Language Model based Autonomous Agents” #)

The advent of Large Language Models (LLMs) marks a milestone, displaying promise in achieving human-like intelligence through extensive training and parameters. LLMs serve as controllers in constructing autonomous agents, aiming for beyond human-like competencies. Research underscores LLMs' pivotal role in enhancing agent capabilities, depicted in the growth trend of LLM-based autonomous agents (Figure 1: “A Survey on Large Language Model based Autonomous Agents” #).

Amidst the evolving landscape of artificial intelligence, Delysium stands out with its innovative framework designed to support a scalable network of AI agents. The architecture of Delysium is streamlined into two primary layers: the Fundamental Layer and the Blockchain Layer, with the overarching ecosystem fostering growth and collaboration.

The Fundamental Layer, also referred to as the Communication Layer, is the backbone of the network, ensuring scalability and facilitating seamless communication among AI agents. This layer is crucial for the real-time exchange of information and coordination between agents, providing a robust foundation for their operations.

The Blockchain Layer offers a secure and transparent platform for AI agents. This layer enhances governance through consensus mechanisms and resource management, ensuring that agent actions are accountable and aligned with the network's protocols.

The ecosystem of Delysium, which encompasses both layers, is a dynamic environment that promotes the discovery and interaction of diverse AI agents and users. It is designed to support the continuous growth of the AI Agent Network and the community, fostering an inclusive space for innovation and development.

By focusing on these core layers and the ecosystem, Delysium addresses the critical need for networks that can efficiently manage a growing number of AI agents and tasks. The integration of blockchain technology within this framework provides the added benefits of enhanced security, transparency, and consensus-based governance, which are essential for maintaining alignment with human values and objectives.

This whitepaper delves into the intricacies of Delysium's architecture, exploring the capabilities and potential of the AI-Agent network and the strategic integration of blockchain technology. It aims to provide a clear understanding of how Delysium's framework is poised to advance AI-Agent networks, driving them towards greater effectiveness and expansion.