No messages
for you.
snazi — short for Soup Nazi AI — guards the inbox your AI assistant reads. Your agent can see who messaged you, but it can't read what they said until you approve that sender — so a stranger can never slip instructions to your AI, and your private messages stay private.
Stores zero message content. An approve / deny list only.
- AllowedMom+1 555 0182
- AllowedRiley (work)riley@acme.co
Your agent sees who. You decide what it can read.
snazi sits between your messages and your AI assistant. Your agent can always see who reached out, but it can only open a message once you've approved that sender.
A message comes in
Someone texts or emails you. snazi records who it's from — but not a word of what they said.
snazi checks your list
If you've approved that sender, your agent can read the message. If not, it stays sealed.
You approve in one tap
snazi sends you a one-tap link to allow or block each new sender. Until you allow them: no messages for you.
Always know who
Your agent always sees the sender's name or address, so you never miss that someone reached out.
Never leak what
Message content stays private until you approve the sender. snazi never stores a single message.
One-tap approvals
Approve or block a new sender straight from a signed link — no logging in, no friction.
Read-only token
Your agent gets a token that can check the list and read approved messages — but can never approve a sender.
Built for every channel
One gate in front of iMessage and more — a single allow/deny list across the channels you connect.
Private by design
No message archive to breach. snazi only ever knows the approve/deny list you control.
We can't leak what we never store.
snazi keeps an approve/deny list and nothing more. No message bodies, no archive, no copies. The gate decides; your conversations stay yours.
Put a bouncer on your agent's inbox.
Set up your list in minutes. Approve the people you trust — ignore everyone else.