It's not a long list.

The organizations in our data that are actually governing AI run three things, attached at the route level, on the path production traffic actually takes:

• Prompt-injection guards: block adversarial inputs before they reach the model
• Request transformers: control what goes out, to which model, under which policy
• Response transformers: control what comes back before it hits your systems

That's it. One control plane. Every model, every provider, every team's requests governed consistently.

Download The State of AI Governance for the full picture: provider market share, multi-model adoption trends, model usage by industry, and what agentic AI means for your governance surface.