Trust Center | Flannel Security & Compliance

Cloud-Native SaaS Platform

Fully managed SaaS hosted on Google Cloud Platform, with infrastructure provisioned and version-controlled through Terraform for reproducible, auditable deployments.
Strict tenant isolation enforced at the application, storage, and catalog layers — every record, query, and workflow execution is scoped to a single tenant by design.
Real-time monitoring of cluster health, query performance, and authentication events, with metrics and structured logs aggregated for alerting and review.

Documented information security policies governing data handling, access management, credential storage, secure development, and incident response.
Policies reviewed regularly and updated to reflect emerging threats, platform changes, and evolving best practices.
Security-aware engineering culture: every team member is responsible for upholding the controls described in this document.

All customer data encrypted at rest with AES-256 and in transit via TLS 1.2 or 1.3, including data stored in object storage, the application database, and connector credential stores.
Custom VPC architecture with dedicated subnets, private database connectivity, and no public endpoints exposed for primary data stores.
Inbound traffic restricted to HTTPS through a managed load balancer; internal service-to-service communication uses native cluster encryption.
Connector credentials and platform secrets are stored in a managed secret manager with envelope encryption, and never committed to source control or written to environment variables at rest.

Identity managed through a trusted third-party identity provider with server-side token verification on every request.
Programmatic access supported via scoped API keys, ID tokens, and standards-based OAuth 2.0 (authorization code with PKCE, client credentials, and refresh tokens).
Two-layer authorization model: role-based access control at the tenant level, plus resource-level policies covering owners, admins, writers, executors, and readers.
Configurable row-level security enforced at the query layer — applied consistently across user queries, workflows, and AI agent tool invocations, with no bypass paths.

Your data powers your notebooks, workflows, and agents — and nothing else. We never sell, share, or monetize customer data.
AI inference uses your data in context only. Your data is not used to train or fine-tune any models, by Flannel or by any underlying model provider.
Only the data relevant to a specific query or workflow step is sent to a model provider, over TLS-encrypted connections, with no caching or retention beyond the immediate request.
AI agent activity is governed by the same authentication, authorization, and row-level security controls as human users — agents cannot exceed the permissions of the identity that invoked them.

Code changes require peer review prior to merge, with automated build, test, and deployment pipelines.
Application dependencies managed through locked dependency files, with vulnerability scanning on the roadmap as part of our compliance program.
Third-party penetration testing and a formal SOC 2 audit are part of our compliance roadmap.
Sensitive workloads run in sandboxed containers on a dedicated cluster, providing kernel-level isolation for untrusted code paths.

Documented incident response procedures covering detection, containment, customer notification, and post-incident review.
Affected customers are notified promptly upon confirmation of any security incident, with details on scope, impact, and mitigation.
No public disclosure is made to third parties without prior written consent from the affected customer.
Customer data is backed up using managed cloud-native backup capabilities, with redundancy across availability zones for primary data stores.

Built on managed Google Cloud services with built-in regional redundancy and managed failover for compute, storage, and database tiers.
Rolling deployments with liveness and readiness probes on all services, allowing platform updates with no scheduled downtime in normal operation.
Continuous health and performance monitoring across application and infrastructure layers, with alerting tied to on-call response.

Customers can export their data at any time via the platform's data export and connector interfaces.
Upon request following termination, Flannel will fully delete customer data and provide written confirmation that all copies have been removed from our systems within 30 days.