Verification and identity
Verify before use (tool authorization)
Verify before use is tool-level authorization for AI agents: mark any tool as sensitive and, before it runs, the agent asks the person for the word of the day, a passphrase, an authenticator code or a phone approval. The answer never reaches the model.
The problem
An agent that can change bank details or read medical records will do it for whoever asks convincingly. Authorization has to happen at the tool, not in the prompt, and the proof must never pass through the model where it could leak.
How verify before use (tool authorization) works
Turn it on for the tool
Open the tool, find Verify before use, and turn it on.
Choose the methods
List methods on the tool and the person must pass each in turn. List none and the company’s methods apply, where any one is enough.
Set how long it lasts
Optionally set Verify again after, such as 5 minutes for a tool that changes bank details. The default is 30 minutes, up to 24 hours.
The agent asks at the right moment
When the tool is about to run, the agent pauses and asks. The answer goes straight to the check and is replaced in history with “[verification answer hidden]”.
Every setting
Who can do what
Tool authors
Turn on Verify before use and choose methods. It’s enforced even in Monitor mode.
Admins
Choose the company’s allowed methods, attempts and lockout, and manage passphrases.
Members
Set up an authenticator and register the phone number they call from.
What to know
- Verify before use is always enforced, even when the policy is in Monitor or the agent’s firewall is Off. It’s a requirement set by the tool’s author.
- Authenticator codes and phone approvals belong to a person. On a call, the caller is matched to a member only when the carrier verified the caller ID, so a spoofed number can’t use them.
- Lockouts apply to the conversation only, never to the member’s account, so an attacker can’t lock your staff out.
- Verification ends early if the voice score rises past the threshold, or if a new person joins the call.
More in Verification and identity
FAQ
Verify before use (tool authorization): questions
How do I require authentication before an AI agent runs a tool?
Turn on Verify before use on the tool and choose the methods. The agent then asks the person to verify before that tool runs, every time the verification window has expired.
Can I require two methods at once?
Yes. List several methods on the tool and the person is asked for each in turn and must pass all of them, for example the word of the day and an authenticator code.
Does the AI see the code or passphrase?
No. The answer goes straight to the check and never reaches the model, logs or transcripts. On calls, codes are typed on the keypad so the AI never hears them.
Can attackers lock out my staff by guessing wrong?
No. A lockout applies only to that conversation for 15 minutes, never to the member’s account or authenticator.
Turn on verify before use (tool authorization).
Turn on Verify before use on any tool and the agent pauses to ask for the word of the day, a passphrase, a TOTP code or a push approval, one or all of them, before the tool runs. Start in Monitor and see what it would catch on your real traffic.
Starts in Monitor · nothing changes for your customers until you enforce
100% customizable, code and all
Make Autessa AI Agent Firewall fully your own.
Use it out of the box, then change anything. Ask Autobot, Autessa’s AI builder, for a new field, screen, report, workflow or agent and it builds it into your copy of the app. Or open the code and edit it yourself. Nothing is locked: the app, its data and its code are yours.
Start free and make it yours