Overview

PrysmAI is the control layer for production AI.

It gives teams one place to route, inspect, govern, and defend AI systems in production.

PrysmAI has two integration surfaces:

  • Proxy path: your application sends model traffic through Prysm
  • MCP path: your agent runtime connects to Prysm through MCP and reports governance events to the same backend

Both feed the same control plane:

  • requests and traces
  • security findings
  • governance sessions
  • policy violations
  • audit evidence

What You See In PrysmAI

The dashboard is organized around the operator workflow:

  • Activity: investigate requests, sessions, and threats
  • Policies: define guardrails, templates, and review violations
  • API Keys: generate runtime credentials
  • Billing: review usage and plan state
  • Settings: configure project and org behavior
Getting Started →