Copilot Studio, Power Automate, Dataverse, PowerApps (Model Driven Apps), Entra ID, Model-Driven App
Copilot Studio
Automate
Dataverse
Teams
Azure OpenAI
Entra ID
Power Apps
RBAC
I built a Dataverse-backed access request agent that captures structured requests in Copilot Studio, applies validation and policy gating, generates a concise AI summary through a reusable BYOM child flow, and creates a tracked request record. From there, watcher flows handle Teams approval, fulfillment, reminders and escalation, while a model-driven app gives operations and governance teams clear visibility into status, traceability and errors.
I built a Dataverse-backed access request agent that captures structured requests in Copilot Studio, applies validation and policy gating, generates a concise AI summary through a reusable BYOM child flow, and creates a tracked request record.
From there, watcher flows handle Teams approval, fulfillment, reminders and escalation, while a model-driven app gives operations and governance teams clear visibility into status, traceability and errors.
This solution is designed around a clean separation of responsibilities. Copilot Studio handles conversational intake and validation, Entra ID supports authenticated identity capture in the secure variant, Power Automate orchestrates approvals and lifecycle automation, Teams provides the approval surface, Dataverse acts as the system of record, and a model-driven app provides operational visibility across the case lifecycle.
These artefacts show how the solution works in practice: conversational intake, approval orchestration, operational visibility, and governed monitoring from the Dataverse record outward. Each component was designed to demonstrate not only functionality, but platform discipline.
The request topic collects application name, access level, justification, duration and approver details, while applying validation rules and an admin-access policy gate before the request is committed.
An approval watcher sends the Teams card once per pending request and writes the decision back to Dataverse. A separate resolution watcher then updates final status, sends the outcome email and logs processing errors.
Once the case enters the correct state, downstream watchers handle approval outcomes, fulfillment updates, reminders and escalation events without relying on manual chasing.
The model-driven app gives approvers and admins a structured operational view of request status, decisions, reminders, escalation activity, summaries, correlation IDs and error details from one place.
This project was designed to demonstrate governance thinking as much as workflow automation. In addition to the happy path, I implemented role-based access, policy-aware validation, identity-backed request capture in the secure variant, duplicate-prevention flags, traceability fields, and resilient TRY/CATCH handling so the workflow is safer to operate and easier to support.
The video below provides a focused walkthrough of the security and governance layer, including identity, RBAC, auditability, idempotency and error handling.
Conversational request intake via Copilot Studio
Reusable AI-generated case summaries (BYOM)
Interactive Teams Adaptive Card approval experience
Dataverse-based lifecycle and status tracking
Autonomous SLA reminders and escalation logic
End-to-end request to fulfillment workflow
Used Power Automate over Agent Flows for approvals to ensure governance and reliability.
Role-Based Access Control (RBAC) implementation
Draft-only editing guardrails and field-level security
Idempotency flags to prevent duplicate processing
Correlation IDs for cross-system tracking
Environment variable-driven configuration
Comprehensive error logging and audit-friendly status tracking
Approval logic separated from the chat layer to improve control, auditability, and supportability.
This project demonstrates how I design Power Platform solutions that balance user experience, process control and operational governance.
Structured conversational intake with guided questioning, validation and policy-aware branching.
Event-driven flows coordinating approvals, reminders, fulfillment and reusable child-flow logic.
A governed data model supporting request lifecycle tracking, status control, summaries and auditability.
Secure request handling with authenticated identity capture and a clear distinction between public and enterprise-ready patterns.
Role-based access, duplicate-prevention logic, traceability fields and resilient error handling built into the solution design.
Building this project reinforced that good business applications need more structure behind the scenes, not less. To make the request experience feel simple, I had to be deliberate about validation, lifecycle state, retry-safe automation and operational visibility. Creating the secure Entra ID variant made that even clearer by showing how the same pattern can move from a portfolio-friendly public demo to a more enterprise-ready authenticated design.
The current case study is deliberately scoped as a portfolio-ready implementation, but the design already supports a broader enterprise direction. The core architecture can be extended without reworking the foundation because the process, data model and controls are already in place.
Expand approval and support interactions inside Teams with richer adaptive experiences, additional approval routes and stronger operational messaging.
Extend the reminder watcher into a fuller service-management pattern with thresholds by request type, escalation paths and more formal queue ownership.
Develop the reusable BYOM child flow further to support richer summarisation, categorisation or downstream triage patterns across multiple business processes.
For full delivery detail, these documents provide the deeper project, governance, and release view.
End-to-end project narrative covering business problem, architecture, data model, workflows, governance, and testing.
Download PDFDeployment-focused document covering solution packaging, environment setup, validation, release controls, and operational readiness.
Download PDF
See how I applied the same Power Platform approach to IT support, combining conversational intake, AI summarisation, escalation logic and operational visibility.