Fanfusion Hub
Trust & Security

Trust by design.

Security is a product feature at Fanfusion Hub, not a procurement artifact. Isolation, approvals, audit, and encryption ship on every squad from day one.

Core controls

Six controls that change the risk profile.

Tenant isolation (RLS)

Postgres row-level security policies scope every query to the current organization. No accidental leaks between tenants — the database itself rejects mismatched reads.

Human-in-the-loop approvals

Sensitive actions route through an approval queue. Operators see the proposed action, the full context, and the policy verdict before green-lighting. Decisions are signed and logged.

Immutable audit trail

Every decision and action appends to an HMAC-chained audit log. Tampering is detectable by design; export is one click for post-mortems or compliance.

Encryption at rest

Managed Postgres with AES-256 at rest. Object storage with server-side encryption. Secrets stored in a dedicated vault, rotated on schedule, and never kept in the application DB.

Safe-launch mode

New squads default to sandbox. Synthetic data, simulated channels, staged approvals — promoted to production only when KPIs hit acceptance thresholds.

Kill-switch & auto-pause

Manual kill-switch halts outbound at squad or org level. Auto-pause triggers on deny-rate or error-rate spikes. Every trigger emits a fresh audit event.

Data handling

What happens to your data — in one place.

Transparency beats marketing. Here is the exact path data takes once your agent starts handling a conversation.

  • Inbound message lands at the channel webhook with signature verified
  • Redaction pass scrubs PII patterns before the LLM call
  • Retrieval pulls only from your org-scoped knowledge base (RLS-enforced)
  • Policy engine evaluates the proposed action; verdict is logged
  • If gated, the action enters the approval queue; if allowed, it sends
  • Delivery receipts and any customer response are reconciled and audited
  • Backups run nightly (Postgres + object storage) with restore drills tested
Operational posture

Ship like it's production — because it is.

Observability

Structured logs, distributed request IDs, and integrations with Sentry for errors. Every request is traceable end-to-end across API, worker, and channel adapters.

Incident response

One-click incident export bundles (incident.json + timeline + audit trail + related runbooks) to accelerate post-mortems or customer disclosures.

Rate limiting & abuse

Per-tenant, per-endpoint, per-IP rate limits. Abuse patterns trigger auto-pause and alert rules. Cloudflare hardening guide available for WAF + bot management.

Compliance

What we will and won't claim.

We prefer plain-language honesty over compliance theater. We're not SOC 2 / ISO 27001 certified today. Our controls are designed to align with those frameworks and our controls matrix is available under NDA.

Available on request: DPA (GDPR-aligned) · SCCs for EU data · Subprocessor list · Incident response playbook · Controls matrix · Penetration test summaries (where executed).
FAQ

Security questions we get.

Are you SOC 2 or ISO 27001 certified?

We are not certified today. Our controls are designed to align with SOC 2 and ISO 27001 principles (access control, audit, encryption, incident response, change management). We can share our controls matrix under NDA.

Where is data stored and processed?

Production data lives in a managed Postgres with encryption at rest. Object storage uses S3-compatible providers with server-side encryption. Regional storage preferences (EU/US) are available on enterprise plans.

How is tenant data isolated?

By Postgres row-level security policies scoped to the current organization. Every session sets the org context; every table enforces it. Isolation is enforced at the database layer, not just the API.

How do you handle PII?

A redaction layer scrubs emails, phone numbers, and card-like strings before LLM calls unless explicitly allowed. Customer identifiers are kept in structured columns, not prompts.

What happens if something goes wrong?

Kill-switch halts outbound actions. Auto-pause rules disable squads on anomaly spikes. Incident export bundles (incident.json + timeline + audit trail + runbooks) are generated for post-mortem or compliance review.

Can you sign a DPA?

Yes. A GDPR-aligned Data Processing Addendum is available on request. For EU customers, we also sign Standard Contractual Clauses when applicable.

Need the full controls matrix?

We'll share our security documentation under a mutual NDA. For procurement questions or vendor reviews, contact the team directly.

For responsible disclosure of vulnerabilities, email security@fanfusionhub.com. We respond within two business days and credit reporters when requested.