😀Summary

SAGA is a pioneering computing framework designed to supply globally distributed and optimized computing power and storage for decentralized networks. It achieves this by linking data and computing resource owners with computational processes in need of these resources. Additionally, SAGA introduces a layer of intelligent interoperability that bridges computational processes with physical computing infrastructures. It forms an ecosystem that adeptly utilizes the community's dormant computing resources, integrating them into a worldwide computation network. The SAGA infrastructure facilitates the optimal placement of AI processes, interfaces, and data within the global network and pioneers machine-to-machine payment and data streaming channels—thereby significantly reducing the costs associated with global computing and enabling innovative business processes within the Data Economy and Internet of Things domains.

The SAGA platform is conceived as an exceptionally flexible network that encompasses mobile consumer devices, edge computing, and IoT devices, alongside PCs, servers, and data centers. This allows for seamless interoperability among its elements and the intelligent automation of workflow design. SAGA harnesses Web 3 technologies, serverless container execution, service mesh orchestration, the crypto-economy, and more, aiming to establish the decentralized "world computer."

SAGA originated as a spinoff from the SingularityNET Global AI Marketplace and has been operational since 2017. It has been nurtured through the SingularityNET X-Lab Accelerator program since early 2018. The technological development and initial use cases of the SAGA platform are primarily (but not exclusively) focused on supporting the computational infrastructure of SingularityNET's decentralized AI agents.

The volume of global computing capacities, data, and computer code is growing exponentially. However, due to the historical context of global computing infrastructure development, computing, data, and code are compartmentalized into silos, accessible only to the wealthiest economic and social entities. The substantial economic and social value trapped within these silos and the potential of decentralized technologies to unlock it is increasingly being recognized and addressed. The cost of utilizing currently available computing resources to train state-of-the-art machine learning models is exorbitantly high, often reaching millions of dollars for a single training cycle, making it unaffordable for most researchers, individuals, and SMEs.

SAGA aims to create a framework that provides an interoperability and transactional layer linking the latent computing resources of privately and commercially owned devices, open-source and proprietary code, data sources, and storage into a single, dynamically evolving Global Economy of Decentralized Computing. By enhancing accessibility—of resources to process owners and of processes to resource owners, data to processes, and storage to data—SAGA will unlock fundamentally new capabilities for value creation and reduce operating costs for the computational processes involved. In the fully functional SAGA framework:

• Computing resource owners (including home or office PCs and laptops, IoT, and mobile devices, etc.) will express and share their unique capabilities, making them available to processes in need across various ecosystems.

• Storage resource owners will articulate and share their storage capabilities, offering them to data owners for free or in exchange for tokens, based on individual preferences.

• Computational process owners will specify their unique computational and data requirements, bid for resources and expertise, and also offer or spot-sell their functionalities and capacities across the ecosystem.

• Data owners will disclose descriptions, access restrictions, and privacy considerations of their data, enabling processes to leverage existing data and contribute to its enhancement and growth.

• Each participant, such as individual users or resource owners, will have the ability to specify how the network should utilize their resources or charge convertible cryptographic tokens for the use of each respective resource.

• Participants, including social and business enterprises, will have the opportunity to bid for free or paid use of resources offered through the framework and pay using native SAGA tokens.

SAGA will facilitate self-regulation among three types of network participants: data, compute & AI service providers, consumers of data & compute (utilizing AI, code, and algorithms), and network operations agents. These agents will coordinate the creation, execution, and verification of computational workflows, connecting providers with consumers into ad-hoc service meshes, where each service is potentially provided by a different entity and compensated according to its preferences. They will ensure the correct and timely provision of services from a decentralized peer-to-peer network, confirm the execution of processes, and ensure appropriate compensation for services rendered by each service provider according to the work performed.

In the simplest scenarios, processes operating on SAGA may be straightforward, standalone computing tasks intended for isolated execution. In these cases, SAGA would merely provide the necessary computing power and data sources to complete these tasks within cost constraints. However, these processes may also be part of various decentralized networks, each with its unique features and ecosystems. In such instances, SAGA embodies what is termed a meta marketplace economy (MME)—an operational modality set to become increasingly valuable as the "network of networks" facet of the emerging decentralized economy transitions from concept to reality with the growing adoption and interoperability of decentralized computing networks (e.g., SingularityNET, DAIA

members, and others).

In the most advanced scenario, which SAGA ultimately targets as the foundation of the global economy of decentralized computing processes, network operation agents will connect data & compute providers from different decentralized (and possibly commercial) computing frameworks with AI, code, and algorithms from various entities into service-mesh type ad-hoc workflows. The SAGA governance and tokenomic framework will ensure that a contract for computational work demanded by a consumer from a workflow is fulfilled, and each component of the workflow is compensated according to the work done, agreed prices, and using cryptographic tokens of involved frameworks and computational engines.

The meta marketplace economy of SAGA will be underpinned by network operations agents utilizing a variety of tools, including:

• A proprietary token that ensures seamless interaction and compliance with existing tokens and value exchange mechanisms within the ecosystems supported by the SAGA framework;

• A governance model designed to foster deep collaborations between human and AI agents across a vast, diverse, and globally interconnected infrastructure.

SAGA is crafted to cater to both commercial and societal objectives synergistically. It aims to act as a commercial hub, bolstering an increasing number of enterprises that require highly customizable and dynamically allocated computing resources, along with workflow designs that are optimized for cost-effective distributed computing powering their decentralized applications. Simultaneously, SAGA is committed to contributing to the societal good by launching initiatives and projects where participants of the SAGA ecosystem are encouraged to offer their resources and expertise to address pressing global challenges.