Microsoft Fabric use case
Microsoft Fabric capacity and refresh questions in your own words
Ask Microsoft Fabric admin questions in Claude, ChatGPT or Copilot: capacity units by workspace, throttling, and semantic models that failed refresh.
Short answer
Microsoft Fabric capacity reporting with Nexatron means asking platform questions in your own words, from the Nexatron app or from Claude, ChatGPT and Copilot: which workspaces consume the most capacity units, when throttling happened and which semantic models failed their refresh. Microsoft Fabric is available, with user-level access: Nexatron inherits existing Microsoft Fabric permissions.
Example questions
- Which workspaces consumed the most capacity units yesterday, and did any trigger throttling?
- Which semantic models failed their last scheduled refresh?
- How did capacity unit consumption by item type change week over week this month?
- Which items ran the most background operations during business hours?
- How many items of each type does each workspace hold?
Agree these definitions first
- Consumption
- Capacity units for interactive or background operations, or both, over which window.
- Throttling
- Which throttling stage counts, such as delay or rejection.
- Failed refresh
- The last refresh only, or any refresh in the period.
- Capacity
- Which capacities, and the workspaces assigned to each.
How it works
1. Connect Microsoft Fabric
An administrator connects Microsoft Fabric with Microsoft Entra ID OAuth 2.0. Nexatron reads the Fabric REST APIs for workspaces and items, and Capacity Metrics on capacity unit consumption and throttling.
2. Permissions
Admin questions need a person whose Fabric role covers the capacity and its workspaces. Users sign in with their own Microsoft Entra ID account, and an answer includes only the workspaces and items their roles allow.
3. Agree the definitions
Fabric administrators approve the definitions above once, so consumption means the same thing in every capacity review.
4. What each answer shows
Questions only read Microsoft Fabric. When a word such as “consumption” or “failed” has more than one meaning, the answer says which reading it used and offers the others in one click. It reads Verified only once a named person on your team has reviewed the logic, with who and when shown. A question Nexatron cannot answer under your rules is declined, with the reason and what to do next.
5. Ask where your team works
People ask in the Nexatron app or through Nexatron’s hosted MCP server in Claude, ChatGPT and Microsoft Copilot. Each answer shows the definition, source and query it used.
Try it on your own Microsoft Fabric data
Bring the questions above, or your own, and the report you compare against today. Do not include credentials or confidential records.