GLOSSARY
Tenant isolation
Tenant isolation aims to prevent one customer's data, permissions or processing from becoming accessible to another customer without authorisation. A tenant may represent a company or a separate organisational unit. Applications, storage, search and operational processes must respect that boundary together.
Confidential proposals need consistent separation
If several sales teams share a technical platform, one customer's preparation must not incorporate another customer's confidential records. Separation in the visible interface is insufficient. Background jobs, search indexes, caches and exports also need correct tenant assignment.
Choosing shared or separate resources affects cost and operating effort. It does not replace effective access control. Customers need to understand which boundaries actually apply and how compliance with them is assessed. Unclear responsibilities between application and infrastructure teams make that evidence harder to produce.
Test prohibited access explicitly
Cases should cover users from different tenants and all relevant data paths. Access must remain appropriately restricted when requests vary or processing is delayed. Vector retrieval and stored memory belong within the test scope. An architecture diagram alone does not prove isolation; concrete test results, reviewed permissions and controlled operational changes are necessary evidence.