How AI Workspaces isolate knowledge and actions
· 2 min read
An AI Workspace is the primary isolation unit inside a Qefro organization: knowledge, instructions, Business Tools, and conversations stay scoped to that workspace.
The failure mode we designed against
When Support FAQs, HR policy, and IT runbooks share one retrieval index, two things happen:
- Wrong answers — the model cites the nearest-sounding passage, not the right audience’s policy.
- Wrong blast radius — a public widget can surface internal content, or an internal tool can become callable from a customer channel.
Workspaces make isolation the default. Channels (website widget, WhatsApp, Internal Portal) bind to workspaces you choose — they do not invent a shared mega-index.
Layered model
Organization (tenant)
└── AI Workspace (Support | HR | IT | …)
├── Knowledge (hybrid RAG)
├── Instructions
├── Business Tools → Business Actions
└── Conversations
- Tenant isolates customers from each other.
- Workspace isolates teams and audiences inside a tenant.
See Multi-tenant AI Architecture and What is an AI Workspace?.
Customer vs employee binding
| Audience | Typical workspace | Channel |
|---|---|---|
| External customers | Customer Support | Website widget, WhatsApp |
| Employees | HR / IT / Ops | Internal Portal + RBAC |
Do not share one workspace across both audiences unless the corpus is intentionally identical and tools are safe for both. Details: Customer AI vs Employee AI.
Practical checklist
- Create one workspace per audience or team.
- Ingest only documents that audience is allowed to see.
- Attach Business Tools last — and only tools that belong in that scope.
- Bind the widget
data-workspace-id(or portal grants) deliberately. - Cross-test: ask Support questions against an HR workspace to prove isolation.
Platform reference: AI Workspaces. Security reference: Tenant Isolation.