Communities hold conversations people would not have anywhere else. This page sets out how that data is protected and what limits the agent layer runs under.
Your data
- Your community content stays yours. You can export all of it at any time and delete it whenever you want.
- We do not train shared models on your community content or on data from your connected integrations.
- Data is encrypted in transit and at rest.
- Integration tokens are encrypted and scoped to the minimum each connector needs.
What agents may do
- Agents read only from the sources you connect, and every answer shows which source it used.
- Every integration is read only until you explicitly grant write access, per integration.
- Agents run in suggest, approve or auto mode. In suggest and approve, nothing reaches a member without a human.
- Every agent action is logged with a reason, and any agent can be paused with one switch.
- You can revoke any connection instantly from the dashboard.
Access and accounts
- Single sign on is available on Growth and above, with SCIM and audit logs on Enterprise.
- Roles and permissions are set per group, so a chapter lead sees only their own members.
- Support staff access to a customer account is logged.
To report a vulnerability, write to contact@nblik.com. We will acknowledge within two working days.