- Concept and Goals of the Project
- Economy and Reward Model
- Technical Architecture of Eidon AI
- Infrastructure and Data Collection
- Team, Funding, and Values
Concept and Goals of the Project
Eidon AI is developing a global network for collecting unique multimodal data aimed at training embodied AI systems. These models require information not only about text or images but also about actions, spatial interactions, and motor skills. To achieve this, platform participants perform tasks that record their real-world actions using special devices or smartphones.
This concept makes it possible to create large-scale datasets that centralized companies cannot access. The platform provides access to participants from around the world, including regions with unique cultural and physical characteristics, increasing the diversity of datasets. In addition, Eidon AI emphasizes the ethical aspect, ensuring that data is collected voluntarily and with rewards for participants. This allows the project to build a sustainable reputation and foster trust among both users and AI developers.
Economy and Reward Model
Eidon AI uses a tokenized incentive model that encourages honest and high-quality contributions from participants. The interaction process works as follows:
- Task placement — the client specifies the type of data, budget, and deadlines.
- Execution — network participants collect or label information.
- Quality check — the data undergoes multi-level evaluation.
- Reward payment — after verification, the client transfers payment via blockchain.
In addition to the basic pay-per-task model, the project considers long-term mechanisms for sharing profits from the commercial use of data. This means that participants may receive income not only once but throughout the dataset’s lifecycle. This approach makes the model more sustainable and attractive for long-term contributors. It also helps form an active community motivated to maintain high quality and completeness of the provided information.
Technical Architecture of Eidon AI
The technical foundation of the platform is built on a modular architecture and automated data quality control mechanisms.
| Component | Function |
|---|---|
| Post Quality MemAgent | Evaluates data by quality metrics, including PQ-score and alignment score |
| User Quality MemAgent | Generates participant ratings and reputation |
| Slashing Mechanism | Penalties for violations and inaccurate data |
| Dispute & Guardian System | Arbitration and participant interest protection system |
| VLM evaluation | Content analysis using vision-language models |
This architecture ensures that only high-quality, verified, and useful data enters the network, while the system remains resistant to abuse. MemAgents automate much of the work in filtering and ranking information, reducing the burden on moderators and clients. Anti-fraud and attack protection mechanisms make Eidon AI more reliable than traditional crowdsourcing platforms. In addition, the open API allows third-party developers to create integrations and extensions, increasing the ecosystem’s value.
Infrastructure and Data Collection
The Eidon AI infrastructure includes mobile tasks (Quests), a marketplace for interaction between clients and contractors, labeling services, and a library of ready-made datasets. Tasks allow participants to quickly start collecting data, while the marketplace offers flexible customization of orders for specific needs.
The project is actively developing specialized devices that make data collection more accurate and convenient. Finger-tracking gloves and eye-tracking glasses make it possible to gather unique information unavailable when using standard cameras. Such data is critical for training AI models that need to understand complex motor and visual scenarios. In the future, Eidon AI plans to expand its range of equipment, including robot sensors and devices for collecting tactile information.
Team, Funding, and Values
The project was founded in 2024 by Peter Toth (CEO) and Sam Padilla (Co-Founder). At an early stage, Eidon AI raised $3.5 million in a seed round from Framework Ventures and cyber Fund, confirming the trust of the Web3 industry.
The project’s values include decentralization, transparency, and preserving data ownership rights. Eidon AI aims to create an environment where every participant has equal opportunities regardless of location or technical resources. Commitment to openness and collaboration helps attract partners from research centers, universities, and technology companies. In the long term, the project aims to become the standard in data collection and distribution for the AI industry, offering a safe and fair way to exchange valuable datasets.
