Security & trust

Security and trust

How BundlHQ protects customer data. This page is a public summary, checked against the live system — the full Security and Trust Overview, DPA and supporting evidence are available on request.

Data BundlHQ holds

  • Menu master data: products, combos, modifiers, ingredients, recipes and costs
  • Prices by sales channel and campaign plans
  • Aggregated point-of-sale sales data by item and store, and supply-chain cost data
  • User accounts: name, work email, role, tenant membership

Data BundlHQ does not hold

  • No payment card data — BundlHQ is outside PCI DSS scope
  • No consumer or end-customer personal data
  • No employee HR data beyond user accounts
  • No store operations, ordering or delivery data

Architecture

Web application over HTTPS, nothing installed on customer devices. Every customer is a tenant with an isolated workspace reached at its own subdomain, with row-level security enforcing that a session can only read or write rows belonging to its own tenant — isolation enforced at the data layer, not only in application code. Production is a dedicated environment used only for paying customers, wholly separate from staging.

Access control and authentication

Role-based access with per-screen permissions set by each customer's own administrators. Multi-factor authentication is available to every user and is enforced as a mandatory step-up before high-risk operations regardless of sign-in MFA status. Production access to BundlHQ's own infrastructure is restricted to named engineering staff, protected by MFA.

Data protection

TLS 1.2 or higher in transit; AES-256 at rest for database, storage and backups. Three independent backup layers, including an encrypted off-platform copy held outside the primary hosting account, with an automated monthly restore drill.

AI features

AI features are opt-in per tenant, with three states — off by default, partial (no cost or recipe data), or full. Customers control which state applies at any time, and AI provider data is never used for model training under commercial API terms.

Monitoring and incident response

Application error monitoring and an automated security-event digest, checked against defined thresholds every four hours with push alerting. A documented incident response process aligned to the NIST Cybersecurity Framework, with severity triage and a 72-hour customer notification commitment for a confirmed incident affecting their data.

Compliance

Designed with reference to recognised frameworks across six jurisdictions — not just a US-centric set, since the same handful of ideas (patching discipline, least privilege, multi-factor authentication, tested backups) recur across every major jurisdiction's baseline. Alignment to a framework is never presented as certification.

Subprocessors

ProviderPurposeLocation
SupabaseDatabase, authentication, storage, serverless functionsAWS Tokyo (pre-production); US region for US-residency customers
VercelApplication hosting and delivery (production, from go-live)Global edge; US origin
DigitalOceanApplication hosting (pre-production)Singapore
Amazon Web ServicesUnderlying infrastructure for the aboveAs above
AnthropicAI features, when enabled by the customerUnited States
GitHubSource control and CIUnited States
Microsoft 365Email, internal documents, off-platform backup storeAustralia
ResendTransactional emailUnited States
SentryApplication error monitoringUnited States

Known limitations, stated openly

Questions, a security questionnaire, or the full Security and Trust Overview, DPA and CAIQ Lite questionnaire — get in touch.

Get in touch