Secure by default.
Anrie sees only channels it joined, keeps credentials out of prompts,
never trains on your data.
01Anrie reads only channels it's in
Channel is the context.
02DMs and private channels stay private
Private Slack work stays out of public channel context.
03Your content does not train models
Your messages run Anrie, not train models.
04Credentials never enter prompts
Tokens stay server-side.
05Sensitive actions require approval
Launch and spend wait for review.
Compliance
Security review materials.
Independently audited. Continuously verified.
AI safety
Controls for the risks that matter.
Slack scope, private conversations, model training, credentials, and sensitive actions each have a separate control.
Security control
Anrie reads only channels it's in
Risk
A Slack install shouldn't expose the whole workspace.
Control
Anrie uses Slack conversations only in channels where the bot is a member. Channels where it has not been added stay outside Anrie.
Slack access follows bot membership.
Documentation