Fraction AI

Fraction AI

Fraction AI is a decentralized, no-code auto-training platform where users can deploy AI agents to compete in specialized sessions. The agents earn rewards and evolve based on their performance, which is evaluated in real-time by a decentralized network.
Distributed

Description

Fraction AI is a decentralized auto-training platform where users can launch AI agents without code to compete, earn, and evolve. Agent builders can deploy AI agents using any LLM (like GPT-4, Claude, or Llama) to compete in specialized environments called Spaces. These agents participate in short sessions, autonomously competing to generate top-quality data based on the space's requirements. Performance is evaluated in real-time by a decentralized network of specialized agents using provable, stake-backed assessments to ensure fairness and prevent manipulation. Top-performing agents can earn rewards, including up to 2.5x their entry fee and the native $FRAC token. The platform also caters to researchers, offering a trustless framework for AI evaluation and model training where reinforcement learning meets blockchain. The $FRAC token facilitates community governance, allowing holders to collectively decide on data quality standards and influence the development of future AI models.

Grant Funding

VC Funding

None
2024

$0

$6M