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

ai-foundry-hub — parity gap (validator v2, 2026-05-26)

Loom URL: /items/ai-foundry-hub/new Fabric/Azure reference: https://ai.azure.com (AI Foundry hub overview) Loom screenshot: temp/parity/ai-foundry-hub-loom.png

Phase 4 — live functional check (HTTPS via minted session)

Route Status Notes
/api/foundry/workspace 200 Real workspace aifoundry-csa-loom-eastus2, kind=Hub, Succeeded, PNA=Disabled
/api/foundry/connections 200 0 connections (empty hub but route is wired)
/api/foundry/deployments 200 0 endpoints, 0 deployments (route works, hub empty)
/api/foundry/computes 200 0 computes attached
/api/foundry/datastores 200 0 registered datastores
/api/items/ml-model 200 0 models (route wires Models tab)
/api/items/ml-experiment 200 0 jobs, 0 experiments

All backing routes return 200 with real Azure data. The hub overview tab renders 12 real metadata fields (storage account safoundryhubm56yejezt7bj, KV kv-loom-m56yejezt7bjo, ACR acrloomm56yejezt7bjo, App Insights ai-csa-loom-eastus2, discovery URL). 7 tabs visible (Overview · Connections · Models · Deployments · Computes · Datastores · Jobs). Ribbon has 4 actions (Reload · Open in Azure portal · New connection · New deployment).

Phase 3 — Fabric vs Loom row-by-row

Fabric AI Foundry hub element Loom present? Severity
Title + friendly name + kind badge YES (Hub badge)
Overview metadata grid (workspace, RG, location, storage, KV, ACR, Insights) YES (12 rows)
Connections tab with list + Auth column YES (table headers wired)
Models tab with version count YES (basic table)
Model catalog (/explore/models): searchbox + 7 filters + paginated cards + leaderboards + detail + Deploy YES (built 2026-05-29) — ModelCatalogPanel, real list-models + deployments PUT
Chat playground (/resource/playground/chat): 3-pane Setup/Chat/Config, real chat completions YES (built 2026-05-29) — ChatPlaygroundPanel, real /api/foundry/chat
Images / Audio / Speech playground tiles YES (honest-gate) — PlaygroundsLandingPanel
Deployments tab (online endpoints + deployments) YES (2 sub-tables)
Computes tab with state + VM size YES
Datastores tab YES
Jobs tab (Experiments + recent jobs) YES (2 sub-tables)
"+ New connection" wizard with provider catalog (OpenAI / AI Search / Storage / Custom) NO — button has no click handler wired here MAJOR
"+ New deployment" wizard with model picker + capacity NO — button is dead MAJOR
Cost meter / quota usage bar NO MINOR
Resource Graph quick-link to provisioning state diagram NO COSMETIC

Buttons clicked / functional

  • Reload — not tested individually but each tab uses useLazyFetch so it should re-call BFF (BFF returns 200, so this works)
  • Open in Azure portal — likely deep-link (not verified opened a new window)
  • New connection / New deploymentBROKEN: button visible but has no onClick handler in foundry-hub-editor.tsx ribbon definition

Grade — B-

The hub editor is the most polished of the AI/ML group. Phase 4 ZERO BROKEN on primary read paths — every tab hydrates real data. Phase 3 has 2 MAJOR (dead "New connection" + "New deployment" ribbon buttons) which would normally drop it to C, but the read experience is honest and real enough that B- is fair — the editor PROMISES write but only delivers read. Mark these two ribbon actions as "Coming soon" or wire them, then it's an A-.