LogoLogo
  • Learn
    • Introduction
      • AegisAI: Modular AI-Specific Layer 1 Blockchain
      • The Need for Decentralized AI
      • Features of AegisAI
      • Typical Use Cases
      • Vision: Building a Fair and Open AI Ecosystem
    • Overview of AegisAI
      • Dual-layer Architecture
      • Lifecycle of an AI Task
      • PoS+W Inference Concensys
      • Reward Mechanism
      • Composability
      • Staking-equivalent License
  • How-to Guides
    • Run a Node
    • Develop a DeAI Application
      • Quick Start Guide
      • AegisAI for EVM Developers
      • AegisAI for Solana Developers
      • AegisAI RPC API
      • Build Your First AI dApp
  • Reference
    • AegisAI LLM Inference Specifications
  • Community
    • Twitter
    • Telegram
Powered by GitBook
On this page
  • Safe and Trustless Relayers
  • Schema
  • Online Content
  1. Learn
  2. Overview of AegisAI

Composability

PreviousReward MechanismNextStaking-equivalent License

Last updated 2 months ago

AegisAI is designed for high composability, enabling seamless integration with blockchain networks such as Ethereum and Solana. This composability allows developers to leverage AegisAI’s decentralized AI capabilities within their dApps across various blockchains.

AegisAI further enhances composability by introducing trustless relayers and schema. Additionally, all-knowing oracles are introduced to support the latest online content.

Safe and Trustless Relayers

Relayers are responsible for submitting tasks generated by blockchain-based Decentralized AI (DeAI) applications and relaying the task results back. They can charge fees for their services to these DeAI applications and need to pay gas fees on the AegisAI network.

To ensure the relay process is secure and trustless, AegisAI introduces a unique identifier, the task hash, for each AI inference task. This ensures that tasks with the same hash produce the same result, allowing relayers to easily relay and verify results using normal smart contracts. Therefore, the EVM/SVM Relayer contracts only need to verify if the latest task data is contained in an AegisAI block and the block is confirmed by enough AegisAI validators.

This mechanism ensures that the composability of AegisAI remains secure and trustless, as the integrity of the results is safeguarded by the task hash and the signatures of AegisAI validator nodes, independent of who performs the relay between the original chain and the AegisAI network.

Schema

To make it easier for normal smart contracts to use AI inference results, AegisAI introduces Schema to mapping string inference results into boolean, address, number that can be directly used in smart contracts.

Schema is executed in both recourse nodes and validators. It ensures that AI computations are predictable, consistent, and easily integrable into dApps.

Online Content

AegisAI supports the integration of online content into blockchain applications through a decentralized oracle mechanism(all-knowing oracles). This feature is particularly useful for dApps that need to access and verify internet-based information, such as social media posts or news articles.

The process is as follows:

  1. User Request: A user or dApp calls an oracle contract with a specific URL and an optional prompt to specify the content to retrieve or analyze.

  2. Oracle Nodes: Oracle nodes listen for these requests and independently retrieve the content from the specified URL. They may use AI models, APIs, or other methods to obtain the data.

  3. Response Submission: Each oracle node submits its version of the content to the oracle contract.

  4. Consensus and Finalization: Once enough responses are collected, the last node to submit triggers a process where all responses are sent to the AegisAI network. AegisAI uses its decentralized AI capabilities to analyze and summarize the responses, producing a consensus-based final result.

  5. Result Upload: The final result is encoded and uploaded to the oracle contract, usually via IPFS for decentralized storage.

This mechanism ensures that online content integrated into the blockchain is reliable and represents a consensus from multiple independent sources.