AgentScore is a critical utility in the agentic economy's financial stack. As agents transition from internal enterprise tools to autonomous economic actors, the need for reputation-based access control becomes paramount. AgentScore provides the necessary data to bridge the gap between anonymous on-chain wallets and trusted service providers, enabling "Know Your Agent" (KYA) compliance for businesses.
The company is active in the middle layer of the agent stack, specifically focusing on the interface between agents and the services they consume. By indexing standards like ERC-8004, AgentScore champions a more transparent ecosystem where agent performance and reliability are public and quantifiable. For anyone building agent-native marketplaces or payment systems, AgentScore represents the necessary safety check that allows autonomous spending to scale without human-in-the-loop verification for every transaction.
Agentic AI is moving from simple conversation to economic action. As agents begin to pay for their own API access, settle invoices, and manage on-chain treasury, they encounter a fundamental bottleneck: trust. In a human-to-human transaction, credit scores and identity verification provide a safety net. In an agent-to-agent transaction, the counterparties are often anonymous wallets. AgentScore is a San Francisco-based startup building the reputation infrastructure to solve this, positioning itself as a credit bureau for the AI economy.
The company works by indexing real-time activity across multiple blockchains, including Ethereum, Base, Arbitrum, and Optimism. Rather than relying on self-reported claims or static metadata, AgentScore derives its grades from historical transaction patterns. They specifically monitor activity on the x402 payment network and the ERC-8004 agent registry. By analyzing these data points, the platform calculates a score that businesses can use to determine whether an agent is a legitimate operator or a potential fraud risk.
AgentScore is accessible via a REST API, providing developers with a simple mechanism to check an agent's reputation before processing a transaction. The core endpoint, /v1/reputation, returns a numerical value and a corresponding grade (e.g., a score of 68 resulting in a 'B' grade). The system also provides visibility into the operator's history, allowing users to see how many other agents are managed by the same entity and their collective performance.
The data is structured to support 'Know Your Agent' (KYA) workflows. This is particularly relevant for marketplaces and API providers that want to offer consumption-based pricing to autonomous agents but need to mitigate the risk of non-payment or resource abuse. By checking the AgentScore before granting access, these providers can implement real-time allow/deny decisions based on the agent's proven reliability.
Based in San Francisco and currently emerging from stealth, AgentScore is built by a team with experience in identity resolution, fraud detection, and data infrastructure. They have secured backing from undisclosed tier-1 investors. While the company is still in its early stages, it has already indexed thousands of agents and transactions across seven major blockchain networks, with the highest concentration of activity on Ethereum and Base.
Their business model follows a standard SaaS pattern. A free tier allows for community experimentation and limited read-only requests, while a $100 per month Pro tier introduces the /v1/assess endpoint for active trust decisions and higher rate limits. Custom enterprise plans are available for high-volume users like major payment gateways or large-scale agent marketplaces that require unlimited quotas and dedicated service-level agreements. This pricing structure reflects a strategy to become the default utility for any service that interacts with the growing number of autonomous spenders in the ecosystem.
Trust grades and reputation scores for AI agents based on real-time on-chain transaction history.
AgentScore is hiring.