Security
An AI agent that asks before it acts.
Data stays in your workspace. The agent asks before every risky action. Admins decide where it may act. Here’s exactly how a request moves.
Data flow
From your request to your tools, step by step.
01
In your workspace
Your request becomes a plan of steps inside your workspace. Nothing is shared with other customers or used for training.
02
Before it runs
Every step is checked against roles and approval rules, and anything that creates, changes or sends something waits for a yes. Traffic uses TLS 1.2+.
03
In your tools
The agent acts only in the tools an admin has approved, with the permissions of the person who asked — and logs every step.
Controls
Admin controls for every workspace.
Single sign-on
SAML and OIDC with the identity provider you already use.
SCIM provisioning
Members created and removed automatically (Enterprise).
Role-based access
Owners, admins and members, with per-tool permissions.
Audit log
Every admin and provisioning event, exportable (Enterprise).
Encryption
TLS 1.2+ in transit and AES-256 at rest.
Data residency
Choose where workspace data lives: EU or US (Enterprise).
Retention controls
Keep history for 0, 30 or 90 days — or set your own.
No model training
Your content is never used to train any model.
Compliance
Your audits and reports go here.
This template does not claim any certification. Replace these placeholders with your own verified audits, reports and policies before you publish.
Security audit
Placeholder — add your status
Privacy framework
Placeholder — add your status
Penetration test
Placeholder — add your status
Data processing terms
Placeholder — add your status
Security FAQ
Questions from security teams.
Can a flow run without approval?
Who can see our flows and runs?
Can we restrict which tools Wordloom writes to?
Do you have a security pack for vendor reviews?
Bring your security questionnaire.
We’ll walk your team through the architecture and answer every question in one call.