Check a
Press ⌘⏎ to run · 5 free checks, no signup
For unlimited use, create a free account and call the API directly with your own key.
An AI agent tries to move $500
A risky action — VENZX stops it and asks a human first.
Customer asks
Agent acts
VENZX holds
Human approves
Action runs
A customer opens a chat with your AI support agent.
Checkpoint
A human reviewer's inbox
When VENZX holds an action,
the approval email lands here.
the approval email lands here.
That's VENZX. Safe actions run instantly. Risky ones wait for a human — one click, by email. Forbidden ones never run. Every decision is logged.
Everything above comes from these lines
guard = venzx.guard_for(policy=Policy()
.require_approval("refund_customer") # ask a human
.allow_domains("your-app.com") # block the rest
)Wrap the function your agent calls, and every decision above happens automatically.
Now try it yourself
Send VENZX any agent action — a user message, an AI reply, or a tool call — and watch it decide in milliseconds: allow, block, or pause for a human, all logged. Click an example below, or write your own — 5 free checks, no signup.
Click an example to try it — it types itself in, runs instantly, and shows VENZX’s verdict