FlowDesk lets AI agents act on your work — so security isn't a feature, it's the foundation. Here's exactly how your data is protected and how every agent action is governed. No marketing, just the architecture.
Every workspace is isolated at the database level with row-level security. One tenant can never read another's data — enforced in Postgres, not just the app.
Agent actions are recorded in a SHA-256 hash-chained, HMAC-signed ledger. Any silent edit breaks the chain — provably. Compliance export on demand.
Agents act through scoped, rate-limited, audited keys, checked against a per-workspace constitution + RBAC, with a one-switch kill-switch and human approval for risky actions.
All traffic is TLS in transit. Stored AI keys (BYOK) are AES-256-GCM encrypted at rest and never returned to the browser.
Anonymous access to internal functions is revoked; the API is origin-locked to the official app and abuse-rate-limited. Function search paths are pinned.
Bring your own AI model key so inference runs on your account. Export your data (CSV) or delete it anytime. We never sell data or use it for ads.
We share data only with vendors that help operate FlowDesk, under contract:
| Provider | Purpose | Notes |
|---|---|---|
| Supabase | Database, authentication, storage | Postgres + RLS |
| Cloudflare | Hosting, CDN, edge functions, security | TLS, DDoS, WAF |
| Anthropic | AI processing of submitted text | Transient; BYOK optional |
| Stripe | Payment processing (if you subscribe) | We never see full card details |
We believe in saying what's true, not what sounds good:
Security incidents affecting customer data, to date: none. We'll publish any incident here, including the ones where the answer is "none."
Security questions, disclosures, or an enterprise review: agents@flowdesk.app. Responsible disclosure welcome — see /.well-known/security.txt.