Introduction
Rakis is a Decentralized Verifiable Inference network that runs entirely in the browser. It's a novel approach to distributing and executing AI inference tasks across a peer-to-peer network, without relying on centralized servers or infrastructure.
The primary goal of Rakis is to democratize access to AI capabilities by leveraging the collective computing power of connected devices worldwide. By harnessing the processing resources of individual browsers, Rakis aims to create a scalable and resilient network for AI inference, free from the constraints and limitations imposed by centralized systems.
At its core, Rakis is a permissionless network, meaning anyone with a web browser can join and participate without requiring special permissions or credentials. We belive that a open and decentralized approach not only promotes accessibility but also fosters innovation and collaboration within the AI and Crypto communities.
One of the key challenges Rakis addresses is the deterministic nature of AI models. Unlike traditional computations, executing the same AI model with the same input can produce different results due to the inherent randomness involved in the underlying algorithms. Rakis tackles this challenge by implementing a novel consensus mechanism based on embeddings, enabling a decentralized network to reach agreement on the outputs of AI inference tasks.
Step 1: Join the Network
To participate in Rakis, all you need is a modern web browser that supports WebGPU and WebAssembly. Simply visit the Rakis website, and your browser will automatically become a node in the network, contributing its computing resources to the collective effort.
Step 2: Request Inference Tasks
Any smart contract or a peer once connected, can send inference requests to the network, specifying your desired AI model, input data, and security parameters. The network will distribute your request across multiple nodes, each executing the inference task independently.
Through its novel architecture and consensus mechanism, Rakis opens up new possibilities for AI applications, enabling decentralized and trustless execution of inference tasks without relying on centralized providers. This not only enhances privacy and security but also paves the way for novel use cases, such as self-executing smart contracts, decentralized AI agents, and more.
To learn more about the story behind Rakis, its motivations, and intended goals, check out the Story of Rakis section. For a comprehensive overview of the technical architecture and core components, refer to the Technical Architecture section.