Tenant Isolation is the boundary that says Customer A cannot read, write, or even observe Customer B's data. Hard isolation at the data layer (per-tenant database row scoping plus row-level security), the network layer (VPC and per-tenant routing rules), and the runtime layer (agents, queries, and workflows refuse cross-tenant references by default).
One layer of isolation is one layer of theatre. Nova enforces tenant boundaries at every layer the request touches. Every database row is scoped, every network request is per-tenant routed, every agent and query checks tenant_id before executing. Removing any one layer still leaves the other two as the wall.
The dangerous failure mode is not "Customer A reads Customer B's data". It is "Customer A reads Customer B's data and nobody notices for six months". Nova logs every cross-tenant access attempt, blocks it at the boundary, and pages the platform team if a pattern emerges. Silent leakage is impossible by design.
Tenant isolation is the default for every Nova customer, but for these three profiles it is the deal-breaker compliance and security review will dig into. If you fit one of these, ask sales for the isolation evidence pack.
For SOC 2, ISO 27001, HIPAA, and customer security reviews you need not only that isolation works, but evidence of every place it could have failed. Every cross-tenant attempt (allowed or denied) is written to the per-tenant audit ledger with the actor, code path, attempt context, and outcome. The evidence pack is a one-click export.
Subscribe to Nova AI Ops on YouTube for demos, tutorials, and feature deep-dives.
Tenant Isolation is the wall between them, enforced at three layers and audited on every attempt. Hard separation, default deny, full evidence trail. Built in, not bolted on.