Skip to content
CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

Copilot in CSA Loom

What Fabric does

Per-workload Copilots are embedded in Fabric: Data Engineering / Data Science (notebooks slash commands), Data Factory (pipelines + dataflows), Warehouse (NL→SQL), SQL DB, Real-Time Dashboard (NL→KQL), Power BI report authoring. Runs on Azure OpenAI. Per-user identity scoping. F2+ paid capacity required (no Trial).

CSA Loom parity design — Loom Copilot runtime

The Loom Copilot is the unified Copilot persona in the Loom Console. Built on the extended apps/copilot/ + azure-functions/copilot-chat/ scaffold (per PRP-09 Data Agents — same runtime, different system prompts and tool catalogs per context).

Personas

The same agent runtime serves multiple personas via system-prompt selection:

Context Persona Tool catalog
Setup Wizard (/setup) "loom-deploy-agent" render_bicepparam, submit_deployment, poll_deployment, activate_pim, MCP tools
Console sidebar (all panes) "loom-copilot" NL2SQL, NL2DAX, NL2KQL, doc-search, workspace-search, capacity-check
Notebook embed (slash commands) "notebook-copilot" /explain, /fix, /comments, /optimize
Warehouse pane "warehouse-copilot" NL2SQL, EXPLAIN, optimize-query
Semantic Model pane (v1.1) "dax-copilot" NL2DAX, DAX-explain, optimize-DAX
KQL pane "kql-copilot" NL2KQL, KQL-explain
Activator pane "activator-copilot" rule-author, threshold-suggest
Data Agents pane "agent-config-copilot" example-query-generate, field-description-generate
Admin pane (v1.1) "ops-copilot" capacity-scale, OAP-toggle, workspace-create

Capacity isolation

Per-organization dedicated AOAI deployment ("Loom Copilot Capacity") in Commercial. In Gov where AOAI TPM quotas are tighter, per-DLZ AOAI deployments for blast-radius isolation.

Telemetry + feedback

Reuses the existing azure-functions/copilot-chat/ feedback + backlog mechanism (per temp/fiab-research/07-existing-repo-scope.md — the existing 400+ line function_app.py has all the security + telemetry + feedback already).

Identity (per AMENDMENTS A15)

OBO throughout — every tool call carries the calling user's Entra token. Per-user audit trail.

Per-boundary behavior

Boundary AOAI Orchestration Foundry portal
Commercial ✅ Full catalog Foundry Agent Service ✅
GCC ✅ Full catalog Foundry Agent Service ✅
GCC-High / IL4 ✅ gpt-4o, gpt-4.1, o3-mini, gpt-5.1 in usgovvirginia/usgovarizona MAF + AOAI direct ❌ (use Azure ML Classic Hub)
IL5 (v1.1) Same Gov catalog MAF + AOAI direct ❌

Honest gaps

  • Cross-workload Copilot continuity — Fabric's Copilot remembers context across workloads in some scenarios; Loom Copilot v1 is per-pane scoped (cross-pane continuity is v1.1)
  • Foundry portal Gov — unavailable; orchestration via MAF, not Foundry portal observability
  • Some advanced AOAI features in Gov — Batch API, Content Safety, Realtime API not in Gov audit tables; Loom Copilot doesn't depend on these for core functionality

Forward migration

Loom Copilot agent definitions export to Foundry Agent Service formats. Per-workload Copilots in Fabric replace Loom Copilot slash-commands when Fabric Gov GA arrives.