Security

Security built into the data layer

AgencyGrid handles carrier credentials, commissions, and PHI, so security isn’t a feature bolted on top. It’s built into the data layer: envelope encryption at rest, per-tenant query isolation, append-only audit logging, and role-scoped access.

Insurance agency data is sensitive by nature. Carrier SFTP credentials, agent NPNs, commission figures, and protected health information all pass through the platform, and a leak of any of them is a real problem for the agency and its clients.

We’d rather be precise than make sweeping claims. Below is exactly how AgencyGrid protects that data today: the concrete mechanisms in the stack, described honestly, without overstating where we are.

Encryption at rest

Sensitive tenant data, chief among it the credentials behind carrier SFTP integration, is encrypted at rest using AES-256-GCM envelope encryption. Each secret is encrypted with a per-record data key, which is itself wrapped by a master key.

The master key lives only in deployment environment variables. It is never stored in the database alongside the data it protects, so a database snapshot on its own does not expose encrypted credentials.

  • AES-256-GCM authenticated encryption
  • Envelope scheme with per-record data keys
  • Master key held only in deployment env, never in the DB

Tenant isolation

AgencyGrid is multi-tenant, and isolation is enforced at the query layer rather than left to application discipline. Every tenant-scoped query goes through a single chokepoint that pre-binds your organization’s id, so a query can only ever see its own organization’s rows.

Tenant tables are not queried directly. Because isolation is centralized in one place, there’s no scattered per-query filtering to forget: one agency’s book, agents, and commissions stay invisible to every other agency.

Audit logging and access

Sensitive actions are recorded in an append-only audit log, the backbone of insurance agency compliance. In particular, staff impersonation of a user and access to PHI are written to that log, which is designed so entries are added, never edited or deleted, giving a durable record of who did what.

Access is role-scoped: agents, managers, and administrators see the slice of the agency hierarchy their role permits, and elevated actions like impersonation are both gated and logged.

  • Append-only audit_logs: records added, not rewritten
  • Impersonation and PHI access recorded explicitly
  • Role-scoped access across agents, managers, and admins

Secrets and where we’re headed

Credentials for infrastructure (the encryption master key, database, storage, email, and billing) live in deployment configuration, separate from tenant data. Authentication is handled by a dedicated auth layer rather than hand-rolled.

On honesty: AgencyGrid does not claim formal certifications it hasn’t earned. Independent attestations such as SOC 2 are on the roadmap, not something we assert today. What’s described above is what the platform does now.

How your data is protected

Security you can actually verify

Envelope encryption at rest

AES-256-GCM with per-record data keys; the master key stays in deploy env, not the database.

Query-layer tenant isolation

Every tenant query is bound to your org id at one chokepoint: no cross-tenant leakage.

Append-only audit trail

Impersonation and PHI access are recorded in a log that’s written to, never rewritten.

Role-scoped access

People see only what their role allows, and elevated actions are gated and logged.

FAQ

Common questions

How are carrier SFTP credentials protected?

They’re encrypted at rest with AES-256-GCM envelope encryption: each credential is protected by a per-record data key wrapped by a master key that lives only in deployment environment variables, never in the database itself.

How do you keep one agency’s data separate from another’s?

AgencyGrid is multi-tenant with isolation enforced at the query layer. Every tenant-scoped query passes through a single chokepoint that pre-binds your organization’s id, so queries can only ever read your own organization’s rows.

What gets audited?

Sensitive actions are written to an append-only audit log. Staff impersonation of a user and access to PHI are recorded explicitly, and the log is designed so entries are added rather than edited or deleted.

Are you SOC 2 certified?

Not today, and we won’t claim otherwise. Independent attestations like SOC 2 are on our roadmap. What we can describe precisely is what the platform does now: envelope encryption at rest, query-layer tenant isolation, append-only audit logging, and role-scoped access.

Handle agency data with confidence

See exactly how AgencyGrid encrypts, isolates, and audits the data your agency runs on.