Speed Reading "Crypto+AI" Rising Star Ritual: Raised $25 Million, the Optimal Solution for Connecting Cryptography with AI Computing Power and Models?

Written by: Karen, Foresight News

The emergence of large language model applications like ChatGPT has greatly accelerated the ongoing evolution of the entire AI ecosystem, giving rise to new paradigms and technological innovations. However, as with the rapid development of any new technology, these large language models also face numerous challenges and issues, such as data privacy and misuse, computational integrity, resistance to censorship, as well as challenges related to licensing and centralized APIs, high computational costs, and monopolies by tech giants, not to mention AI governance and ownership.

To address these issues, Ritual has emerged as a project envisioned as an open, modular, sovereign execution layer for AI. Ritual will combine a distributed network of nodes with computational capabilities and model creators, allowing these creators to host their models on the nodes. Users can then access any model on the network (whether LLM or classic ML models) through a universal API, and the network features additional cryptographic infrastructure to ensure computational integrity and privacy.

What is the background of Ritual?

In November 2023, Ritual announced that it had officially emerged from stealth mode after months of development and publicly unveiled its project. Ritual aims to integrate the best principles and technologies of cryptography and AI to create a system that allows for the open and permissionless creation, distribution, and improvement of AI models.

The Ritual website lists 21 team members, including co-founder Niraj Pant, who was a general partner at Polychain and is currently 27 years old. He joined Polychain as an intern at the age of 19 and dropped out of the University of Illinois a month later. Niraj Pant has researched privacy at the Decentralized Systems Lab at the University of Illinois Urbana-Champaign.

Another co-founder, Akilesh Potti, is also a former partner at Polychain and has worked in machine learning, high-frequency trading, and systems building. Other team members include Ricky Moezinia, former chief AI engineer at Microsoft AI, former software engineer at Facebook Novi and Diem (founder of cross-border payment company Alta), as well as members from Polychain, Trust Machines, Dragonfly, Protocol Labs, dYdX, and others.

At the time of the public project announcement, Ritual also announced the completion of a $25 million funding round led by Archetype, with other investors including Accomplice, Robot Ventures, dao5, Accel, Dilectic, Anagram, Avra, and Hypersphere.

Ritual's angel investors include former Coinbase CTO Balaji Srinivasan, Protocol Labs researcher Nicola Greco, Worldcoin research engineer DC Builder, EigenLayer chief strategy officer Calvin Liu, Monad co-founder Keone Hon, and Daniel Shorr and Ryan Cao from the AI+Crypto project Modulus Labs.

Ritual's advisory board is also quite strong, featuring Illia Polosukhin, co-founder of NEAR Protocol and co-author of the groundbreaking research paper "Attention Is All You Need" (Foresight News note: published in 2017, regarded as the foundation of the Transformer language model, and the source of the various GPT models that have emerged since), EigenLayer founder and partner Sreeram Kannan, and Gauntlet founder and CEO Tarun Chitra. Subsequently, BitMEX co-founder Arthur Hayes also joined Ritual as an advisor.

How is the Ritual architecture structured?

Ritual supports the seamless integration of AI into applications or protocols on any chain, and it supports fine-tuning, monetization, and execution reasoning. Ritual states that its ultimate goal is to enable developers to build fully transparent DeFi, self-improving blockchains, autonomous agents, generative content, and more.

Currently, Ritual has launched a lightweight library called Infernet that brings computation on-chain, which is the first phase of Ritual's product. Infernet can also be seen as a decentralized oracle network optimized for AI, supporting any EVM-compatible chain and allowing smart contracts to access various on-chain use cases and tasks' AI models locally.

Specifically, Infernet enables smart contract developers to request off-chain computation through Infernet nodes and pass the computation results back to on-chain smart contracts via the Infernet SDK.

1. Infernet nodes are lightweight off-chain clients of Infernet, responsible for listening to requests and completing computational workloads.

2. The core of the Infernet SDK is the Coordinator, which manages the registration and activation of nodes in the network and allows users to subscribe to the outputs of off-chain computational workloads. Subscriptions are requests made by users to Infernet nodes to process certain computations (one-time or recurring). Users initiate subscriptions, and nodes fulfill these subscriptions.

Infernet states that developers can delegate computation-intensive operations (such as ML inference or ZK proof workloads) to off-chain, consuming outputs and optional proofs in smart contracts through on-chain callbacks.

Regarding whether there will be incentives for Ritual, Ritual states that various participants in its network (including computation providers, model creators, proof providers, etc.) will be incentivized.

Ritual plans to launch its second phase in the coming months, which involves starting its own sovereign chain, Ritual Chain, and will have its own custom virtual machine acting as a co-processor (validating computations) to serve more advanced AI-native applications that will natively exist on the Ritual Chain. On Ritual's sovereign chain, ZK will become a key component for scalability.

Currently, participation in Ritual can be achieved by running Infernet nodes or by being a user of applications or protocols utilizing Infernet.

Ritual collaborates with EigenLayer and io.net to accelerate decentralized AI

In late February, Ritual announced two partnerships, first with EigenLayer, a "leader" in the restaking space, and then with io.net, a "newcomer" in AI computation within the Solana ecosystem. Ritual is developing an AI-native Active Verification Service (AVS) to support various parts of Ritual's Infernet and Ritual Chain while also releasing new AI-native opportunities for operators on EigenLayer. Ritual states that thanks to EigenLayer's high TVL and diverse node set, Ritual Chain and its many features will enjoy a high level of economic security and decentralization from the outset.

The following image is an example of model operations conducted through EigenLayer operators. EigenLayer operators with access to computation can restake and register as Ritual nodes to provide users access to these operations. The Ritual AVS contract acts as a coordinator, providing coordination services after users initiate computation requests. Then, Ritual phases and operators receive computation requests from the coordinator, after which operators retrieve relevant models from model storage, compute, and return inference outputs to the coordinator, which then outputs the results to the users. As Ritual integrates these services' payment flows, restakers will benefit from the resulting revenue.

Additionally, since Ritual can guarantee the computational integrity of model operations, once EigenLayer's Slash mechanism is online, EigenLayer operators registered as nodes on Ritual can begin providing proofs for generated model operations. If they are found to generate erroneous inferences, their posted collateral will face Slash penalties.

Regarding the collaboration with io.net, Ritual will leverage io.net's decentralized GPU stack to power the Ritual network, further accelerating its decentralized AI goals. This means that the Ritual client will be able to access GPUs on io.net locally, allowing users to easily launch models and serve various applications both on-chain and off-chain. Ritual states that with the launch of Ritual Chain, nodes running GPUs will be able to participate in securing the chain and serving AI-related workloads, thereby increasing and diversifying revenue sources for io.net GPU providers.

Ritual states that as users gain access to an increasing number of models and nodes with various functionalities, it will allow users to preferentially route to the best models and best providers. This capability not only enhances the efficiency of its system but also ensures that users receive the highest quality of service. Furthermore, restaked EigenLayer nodes can also act as routers within the Ritual system, creating matching engines on Ritual, further enhancing the system's flexibility and scalability.

Achieving a truly open, secure, and decentralized AI system is no easy task, requiring the resolution of numerous technical and coordination challenges, such as standardization, incentives, governance, and more. Standardization is key to ensuring that different systems and components can work together seamlessly, incentives are crucial for attracting quality node providers and users, and governance ensures the stable operation and sustainable development of the entire system.