Your content. Your data. Your rules. Every action audited, every transition enforced.
Every channel output passes through the safety pipeline before it can be marked Ready. The pipeline blocks the Ready transition if any check fails.
200+ default terms across 8 categories (hate, nsfw, violence, competitor, legal, drugs, spam, custom). L33t-speak normalisation. Custom workspace terms. Fast, synchronous.
Classification only — not generation. Strict system prompt checks for hate speech, NSFW, harmful content, and brand mismatch when a Brand Profile is set. Runs async after save. Result cached on channel output.
Extracts all URLs from content, verifies they resolve with HEAD requests. Flags broken links before publication. No dead URLs going live.
All three must pass → safety_status = passed. Any failure → blocks Ready transition. Owner can override with explicit confirmation (audit logged).
Full access. Manages workspace settings, team members, brand profiles, channel accounts. Can override safety flags.
Create and edit Stacks and channel outputs. Mark Ready. Publish. Cannot manage settings or team members.
Read-only access. Can view Stacks, Queue, and published content. Cannot create or modify.
Workspace scope is always derived from the JWT — never from a URL parameter. Permission boundaries are simple because the product is simple.
Austrian company subject to EU data protection law. EU-hosted database (Frankfurt). Full GDPR compliance with data export and deletion on request. US data transfers (Ayrshare, Vercel, OpenAI) disclosed with SCC reference. No data sharing beyond what is needed to operate the service.