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

Day 4 — BI & AI & Direct Lake & Data Agents (Commercial CoE)

Track: 5-Day Commercial CoE Workshop · Day 4 of 5 · BI & AI & Direct Lake & Data Agents

Day 4 builds a semantic model + report, demonstrates the Direct-Lake-Shim with an honest latency discussion, and authors a Loom Data Agent — on Commercial, optionally integrated with the Foundry Agent Service (GA).

Azure-native by default

Semantic models use the Loom-native tabular layer (Power BI F-SKU optional on Commercial). Data Agents run on Azure OpenAI; on Commercial the Foundry Agent Service is available (GA) for server-side thread persistence. Loom still runs without Fabric — LOOM_DEFAULT_FABRIC_WORKSPACE unset.

Learning objectives

  1. Author a semantic model over Gold + a report.
  2. Configure the Direct-Lake-Shim and reason honestly about latency.
  3. Author + test a Loom Data Agent (optionally on Foundry Agent Service).
  4. Design an Activator rule firing into Teams.
  5. Use Loom Copilot build-assist.

Facilitator guide

Timing (8-hour day)

Time Activity Mode
09:00 Day-3 recap + semantic-layer architecture Lecture
09:30 Author a semantic model over gold.device_hourly Lab
10:30 Break —
10:45 Direct-Lake-Shim config + honest latency demo Lab + Lecture
11:45 Lunch —
12:45 Data Agent authoring + Foundry Agent Service integration Lab
14:00 Activator rule → Teams Lab
15:00 Break —
15:15 Loom Copilot build-assist tour Lab
16:15 Wrap-up + Day-5 readout prep Plenary

Talking points

  • Semantic layer: Loom-native tabular layer over Gold. On Commercial a Power BI F-SKU is available if the customer prefers native Power BI; it is optional, not required.
  • Direct Lake honesty (read verbatim): Fabric Direct Lake refresh is framing — a metadata operation that takes a few seconds, on by default. The Direct-Lake-Shim reproduces this warm-cache pattern on Azure-native by reframing on Event Grid signals; the freshness window is seconds to tens of seconds. On Commercial, customers also have a live path to Direct Lake on OneLake via forward migration (Day 5).
  • Foundry Agent Service (GA on Commercial): Data Agents can use server-side thread persistence + the Agents playground. Contrast with the Federal track, where Foundry Agent Service is Forecasted in Gov.

Exercises

  1. Measure the shim refresh latency live; calibrate expectations honestly.
  2. Each participant configures a Foundry Agent Service thread store and confirms persistence across sessions.

Common pitfalls

  • AOAI deployment missing → the Data Agent pane names the env var/deployment; provision + retry.
  • Treating the shim as sub-second — keep the latency discussion honest.

Participant lab — BI + AI

  1. Semantic model. In Semantic model (/semantic-model), build measures
  2. dimension + date relationship over gold.device_hourly. Save.
  3. Report. Build a timechart + top-N table; confirm it renders on real data.
  4. Direct-Lake-Shim. Configure for the Gold table, push a Gold update, measure the reframe latency. Record for Day-5 readout.
  5. Data Agent. In Data agent (/data-agent), create an agent grounded on the IoT lakehouse. Optionally wire the Foundry Agent Service thread store. Ask: "Which device had the highest average reading in the last 24 hours?"
  6. Activator rule. In Activator (/activator), fire a Teams message on a threshold breach; trigger with a synthetic spike.
  7. Copilot. Use Copilot (/copilot) build-assist to scaffold a measure.

Validation (Day-4 done): semantic model + report on real data, shim reframe measured, Data Agent grounded answer (with Foundry thread persistence if wired), Activator posts to Teams.

Datasets

Homework

  • Prepare Day-5 readout slides incl. the live forward-migration plan.

Commercial-specific emphasis

  • Foundry Agent Service available (GA) — server-side threads + playground.
  • Defender for Cloud AI Threat Protection available on Commercial (no Sentinel workaround needed).
  • Power BI F-SKU optional — native Power BI is an available alternative to the Loom-native tabular layer.

Slide deck

make loom-decks DECK=docs/fiab/workshops/5-day-commercial-coe/day-4-bi-ai.md.