Oracles

Oracles provide the external data required to resolve prediction markets on the Bluff platform. The system is designed around the EventOracle smart contract, which supports two distinct resolution models: an AI-powered oracle and a user-designated oracle.

Both models are managed through the EventOracle contract to ensure a consistent and secure process for market resolution.

For a detailed explanation of the oracle system's architecture, core functions, and the differences between the two models, please see the comprehensive guide:

AI-Powered Oracles

Bluff integrates AI-powered oracles to decentralize and automate market resolution. This system is intended for markets where the outcome can be determined by analyzing publicly available information.

How It Works

The AI oracle system uses a network of independent AI agents to provide decentralized market resolution. The key features of this system include:

  • TEE-Native Operation: Agents operate within a Trusted Execution Environment (TEE) to protect their data processing and decision-making from external tampering.

  • Research Capabilities: Agents can use various tools, such as internet search (DEEP_RESEARCH) and social media access (TWITTER), to gather information. They can also use multiple AI models (e.g., GPT-4o, Claude 3.5 Sonnet) to analyze data.

  • Source Quality Evaluation: Agents are designed to evaluate the quality and reliability of data sources to mitigate the risk of resolutions based on misinformation.

  • On-Chain Consensus: Agents submit their findings, along with a reference to their reasoning (an ipfsHash), to the EventOracle contract. A market is resolved only after a configurable number of agents (iterationsRequired) reach a consensus on an outcome.

Opportunities and Challenges

Opportunities:

  • Scalability: AI oracles can resolve a high volume of markets in parallel.

  • Speed: Automated resolution can lead to faster settlement times.

  • Objectivity: For certain markets, AI may provide an unbiased assessment of factual data.

Challenges:

  • Ambiguity: The system is not well-suited for markets with subjective or ambiguous outcomes.

  • Data Integrity: The system must be robust against manipulated or biased data sources.

  • Security: The AI network must be secured against coordinated attacks.

Rollout Plan

AI oracles will be rolled out in phases, beginning with markets that have simple, binary outcomes and are easily verifiable using publicly available data. As the technology and security measures mature, their use may be expanded to more complex scenarios.

PreviousFeesNextThe Bluff Event Oracle

Last updated