Knowing a real human owns an agent is just the start. The harder question is: did that human actually approve what the agent just did?
AG9 answers that. It's VeryAI's product for real-time agent action authorization. Every sensitive action an agent takes a payment, a publish, a transfer can be tied to a fresh palm scan from the human behind it. Not a stored token. Not a cached session. A live, scoped approval.
How It Works
A partner platform integrates AG9 alongside their existing authentication
A user links their agent by scanning its pairing QR code in the VeryAI app and palm-scanning to confirm
When the agent tries something sensitive, the user gets a prompt on their phone
The user approves or denies it with a palm scan
AG9 issues a signed proof that the human approved that specific action
The platform verifies the proof in under 100ms and allows or blocks the action
💡 To use AG9 you will need your VeryAI account to be active.
Who Is It For?
Platforms that need proof a human approved a specific agent action, not just that an agent exists
Developers building agents that interact with sensitive APIs, financial services, or regulated systems
Enterprises that need an auditable record of which human authorized which agent action
If you're a regular VeryAI user you don't need to set anything up. When a platform you use requires action approval, you'll simply get a prompt in your VeryAI app.
How AG9 Relates to ClawKey
ClawKey proves a real human owns an agent. AG9 proves that human approved a specific action. They work together as the full picture of agent accountability.
Get Started
🌐 ag9.ai
📖 Integration Guide: ag9.ai/integrations