Building a Reputation Layer for AI Agents on NEAR Protocol

Research context and background

As of 2026, AI agents are actively handling real money. These agents power trading bots, DeFi automation, and analytics tools. However, a critical question remains unanswered: Who vouches for them? Currently, users encounter AI agents with landing pages, performance claims, and Telegram groups, but zero verifiable history. There are no reviews tied to real wallets, and no mechanism to confirm if anyone has actually used the service. This environment resembles Fiverr without ratings or Uber without driver scores, forcing users to rely entirely on trusting vibes rather than data.

Benefits

B7systems solves this trust gap by creating an on-chain reputation layer for AI agents built on the NEAR Protocol. The platform ensures authenticity through a straightforward loop. First, an AI agent registers on the platform. Next, users leave reviews tied directly to their NEAR wallet addresses. Then, a reputation score from 0 to 100 is calculated based on these verified reviews. Finally, agents can achieve on-chain verification via a smart contract. By linking every review to a specific NEAR wallet address, the system prevents spam from fake accounts, creating a transparent trust mechanism.

The project utilizes a robust tech stack to support both current manual registration and future autonomous capabilities. The frontend is built with Next.js, the backend API uses FastAPI in Python, and the database is Supabase. The smart contract is deployed on the reputation-oracle.testnet, and the hosting is on Vercel. The most innovative aspect of this build is the utilization of NEAR chain signatures. This technology enables a unique capability: AI agents owning their own wallets and registering themselves autonomously. Currently, developers manually register their agents, but the infrastructure is already in place for them to register themselves without human intervention in the future.

Use Cases

B7systems is designed for anyone building or using AI agents that handle real money or critical tasks. It is particularly useful for users of trading bots, DeFi automation tools, and analytics services who need to verify the reliability of an agent before interacting with it. The platform currently has 13 registered agents and 30 reviews submitted, with 10 agents currently verified. A public REST API allows anyone to pull reputation data, making it easy for developers to integrate trust scores into their own applications. The system is also seeking early users who have built an AI agent and want to register it on the platform.

Pricing

No pricing details are available for B7systems at this time.

Vibes

No specific reviews, testimonials, or public reception details are available beyond the current adoption metrics of 13 registered agents and 30 reviews.

Additional Information

The project is actively seeking early users. If you have built an AI agent, you can register it at b7systems.vercel.app/register, a process that takes approximately two minutes. Additionally, the team is looking for co-founders interested in the intersection of AI and Web3 infrastructure. The project's source code is open on GitHub at github.com/Zhavkh/ORACLE.