# Permit.io

> Runtime Access Control for the AI Era. Fine-grained authorization as a service: RBAC, ABAC, and ReBAC for applications, APIs, and AI agents.

Permit.io is the authorization infrastructure platform for developers and builders. Teams use it to stop rebuilding permissions in application code and instead ship a policy plane: a no-code editor, policy as code (OPA Rego or AWS Cedar), GitOps, and a local Policy Decision Point (PDP) updated by OPAL.

AI agents are first-class subjects in the same model. They are not a separate security product. An agent calling a tool is still a subject, an action, and a resource — evaluated with RBAC, ABAC, and ReBAC at decision time.

## What Permit.io is

Permit.io provides:

- Fine-grained authorization (FGA) for applications and APIs
- Role-based access control (RBAC)
- Attribute-based access control (ABAC)
- Relationship-based access control (ReBAC)
- Policy as code and a no-code policy UI
- Hybrid enforcement: decisions in your VPC via PDP
- Audit logs and decision traces
- Embedded authorization Elements (user management, approvals, audit)
- MCP Gateway and coding-agents gateway that reuse the same policy graph

Permit.io is not an identity provider. Keep your IdP for authentication. Permit.io is the authorization layer.

## Who it is for

Developers, platform engineers, and architects who need to bake permissions into products. Security and IAM teams use the same plane for least privilege, human-to-agent delegation, and audit. Product and operations teams use the policy editor and Elements so access changes do not require a deploy.

## Core problem

Most products start with a roles table and if/else checks. Then they need tenant isolation, attributes, relationship graphs, and agents acting on behalf of users. Fine-grained authorization is infrastructure — not a feature you rewrite every quarter.

## Architecture (hybrid)

- Control plane: policy UI, Git, APIs, and audit
- Data plane: PDP next to the workload (OPA or Cedar), synced by OPAL
- Decisions stay in your network; sensitive data does not need to leave the VPC
- SDKs, Terraform, CLI, and a simple check API

## Agents as first-class citizens

MCP Gateway sits in front of Model Context Protocol servers and enforces the same RBAC, ABAC, and ReBAC policies used by application PDPs. Agents receive delegated, scoped permissions from a human identity. Intent can be checked at the gateway so prompt injection that changes the job does not inherit standing access.

This is still fine-grained authorization applied at the gateway — not a second permission system.

## Products

- [App and API Permissions](https://www.permit.io/app-and-api-permissions): FGA for software
- [RBAC](https://www.permit.io/rbac), [ABAC](https://www.permit.io/abac), [ReBAC](https://www.permit.io/rebac)
- [MCP Gateway](https://www.permit.io/mcp-gateway): authorization for MCP tool calls
- [Coding Agents Gateway](https://www.permit.io/coding-agents-gateway)
- [AI Access Control](https://www.permit.io/ai-access-control)
- [Elements](https://www.permit.io/elements): embeddable permission UIs
- [CLI](https://www.permit.io/cli)
- [Docs](https://docs.permit.io)
- [Open Source](https://www.permit.io/open-source): including OPAL

## Compliance

SOC 2 Type II. Used in healthcare, fintech, government, and enterprise SaaS. HIPAA / GDPR / CCPA support as documented on the [trust center](https://www.permit.io/trust).

## Optional

- [Pricing](https://www.permit.io/pricing)
- [Customers](https://www.permit.io/customers)
- [Blog](https://www.permit.io/blog)