What is Mandaitor?
Mandaitor is a neutral, trustworthy infrastructure for verifiable delegated authority between humans and AI agents. It provides the tools to safely and transparently grant specific permissions to automated systems, ensuring they operate only within their intended boundaries.
In a world increasingly reliant on AI, Mandaitor acts as a digital notary, creating a secure and auditable record of who can do what on whose behalf. This solves a critical problem: how do you trust an AI agent to act for you without giving it unlimited access?
The Core Problem
Imagine you want an AI agent to manage your construction projects. You want it to approve invoices, but only up to a certain amount. You want it to validate building plans, but not to sign legally binding contracts. Without a system like Mandaitor, you would have to build complex, custom permission logic into every application.
Mandaitor externalizes this logic into a centralized, verifiable registry.
How It Works
Mandaitor is built on three pillars:
-
Mandates: A digital agreement that specifies the Principal (who is delegating), the Delegate (the AI agent receiving authority), the Scope (what actions are allowed on which resources), and the Constraints (the rules and limits).
-
Verification: A high-speed, low-latency API endpoint (
/verify) that checks in real-time whether an action proposed by an agent is permitted by an active mandate. -
Proof-of-Mandate: A cryptographically signed W3C Verifiable Credential (VC) that serves as irrefutable proof that a specific action was authorized at a specific time. This enables non-repudiation and builds trust between all parties.
By combining these elements, Mandaitor provides the foundational infrastructure for a future of secure and scalable AI-to-human and human-to-AI collaboration.