Back to showcase

Reference build / Agent workflow pattern

Fabric Data Agent Workflow Pattern

A governed conversational analytics and workflow pattern based on Fabric Data Agent concepts and Microsoft Fabric Data Science guidance.

04. AI Agents & Workflow AutomationSynthetic governed analytics workspace scenarioStatus: guildbuild adaptedLast reviewed: 2026-05
Fabric Data Agent Workflow Pattern official-source architecture visual
Official Microsoft source visual used with attribution; GuildBuild adds the delivery checklist and evaluation lens below.

Official resources used

GuildBuild adaptation

GuildBuild adapts Fabric Data Agent concepts into an operating pattern for governed analytics Q&A, escalation, and audit review.

What Microsoft provides

  • Fabric Data Agent concepts for natural-language Q&A over governed Fabric data.
  • Data Science documentation for Fabric AI and ML workflows.
  • Platform context for lakehouse, warehouse, semantic model, and KQL sources.

What GuildBuild adds

  • Question intake, permission, escalation, and response-quality review checklist.
  • Human-review path for sensitive or high-impact analytics questions.
  • Agent operating model that keeps the production API/model choice separate from this showcase.

Buyer evaluation checklist

  • Do users know which governed sources the agent is allowed to query?
  • Are sensitive questions escalated before answers are acted on?
  • Can admins review questions, answers, and quality issues over time?

Implementation deliverables

  • Agent source-readiness checklist
  • Prompt and instruction review worksheet
  • Escalation and audit flow
  • Demo script for governed analytics Q&A

Technologies shown

Fabric Data AgentFabric Data ScienceOneLakePower BI semantic modelsKQL databases
Related GuildBuild service: 04. AI Agents & Workflow Automation