Employee Reimbursement Automation System

MICROSOFT POWER AUTOMATE FINAL PROJECT Employee Expense Reimbursement Automation Final Training Assignment | Completion Window: 20th Aug 2026 1. Project Objective Build end-to-end employee reimbursement automation using Microsoft Forms, Power Automate, Dataverse, Outlook, AI Builder Prompt, AI Builder Document Processing, Approvals, and reusable Child Flows. The solution should receive reimbursement requests, validate the submission, process the supporting document, use AI to evaluate the request, route it through the appropriate approval process, update Dataverse throughout its lifecycle, and notify relevant users. Excel must not be used as part of the solution. Supporting documents may include PDF, JPG, PNG, or DOCX files, but Excel files must not be processed. 2. Employee Request Form Create a Microsoft Form named Employee Reimbursement Request. Employee Name Employee Email Employee ID Expense Category Expense Date Claimed Amount Currency Business Justification Manager Email Supporting Document Expense categories: Travel, Hotel, Meals, Transportation, Training, Other. The employee must be able to upload a supporting document. 3. Dataverse Requirements Create a Dataverse table for reimbursement requests containing, at minimum: Request ID Employee Name Employee Email Employee ID Expense Category Expense Date Claimed Amount Currency Business Justification Manager Email AI Validation Result AI Confidence Score Status Rejection Reason Submitted Date Create a second table for processed documents containing: Document ID Request ID File Name Document Type Vendor Document Date Extracted Amount Confidence Score Processing Status Create a third table for Approval History containing, at minimum: Approval ID Request ID Approver Decision Comments Approval Date 4. Request Validation Validate each new request before normal processing. The solution must identify: Missing Employee ID Missing Manager Email Zero or negative claimed amount Expense date in the future Missing supporting document Invalid requests must be recorded in Dataverse with an appropriate status and the employee must be notified through Outlook. 5. Duplicate Detection Check Dataverse for an existing request where Employee ID, Expense Date, and Claimed Amount match the new request. Duplicate requests must not proceed through normal processing. Record the duplicate status and notify the employee. 6. Document Processing Process the uploaded supporting document using AI Builder Document Processing. Extract relevant information such as: Vendor name Document/receipt number, if available Document date Total amount Currency Document type Store the extracted information in Dataverse and compare the extracted amount with the employee's claimed amount. 7. AI Builder Prompt Create an AI Builder Prompt to evaluate the reimbursement request. The prompt should consider: Business justification Expense category Claimed amount Extracted amount Vendor Document date Document type The prompt should determine: Whether the document appears relevant to the request Whether the claimed amount matches the document Whether the expense category is appropriate Whether there are inconsistencies Risk level Recommendation Confidence score Reason for the recommendation The prompt output must be structured so that Power Automate can reliably consume it. 8. Business Rules Expense age: If the expense is more than 60 days old, automatically reject the request. Amount: If the claimed amount is ₹5,000 or less, require manager approval. If it is greater than ₹5,000, require manager approval followed by Finance approval. Amount mismatch: If the claimed amount does not match the extracted document amount, send the request for manual review. AI confidence: If AI validation confidence is below 80%, send the request for manual review. Successful validation: If validation passes, proceed to the appropriate approval process. 9. Approval Use Power Automate Approvals. Requests of ₹5,000 or less require Manager approval. Requests above ₹5,000 require Manager approval followed by Finance approval. Approvers should be able to see employee details, category, date, claimed amount, extracted amount, AI validation result, AI confidence, and business justification. 10. Dataverse Status Management Maintain an appropriate status throughout the request lifecycle. Suggested statuses: Submitted Validation Failed Processing Manual Review Awaiting Manager Approval Awaiting Finance Approval Approved Rejected Duplicate Failed Update the Dataverse record whenever the request moves to a new stage. 11. Outlook Notifications Submission confirmation to the employee Validation failure notification to the employee Manual review notification to the employee Approval request notification to the manager Finance approval notification where applicable Approval confirmation to the employee Rejection notification to the employee with the reason where available 12. Child Flows The solution must contain at least two reusable Child Flows. Child Flows should have clearly defined inputs and outputs and should provide reusable functionality rather than simply splitting one large flow. 13. Error Handling Implement proper error handling for failures involving Dataverse, Document Processing, AI Prompt, Outlook, or Approvals. Update the Dataverse record with an appropriate failure status Capture useful error information Notify the support/automation team Avoid leaving requests stuck indefinitely in an intermediate state 14. Solution Management The entire project must be created inside a Power Platform Solution. Use appropriate Connection References Use Environment Variables where appropriate Follow consistent naming conventions Avoid hard-coding configurable values such as email addresses where an environment variable or configuration approach is appropriate 15. Testing Requirements Test at least the following scenarios: Valid request under ₹5,000 Valid request above ₹5,000 Missing mandatory information Missing attachment Duplicate request Expense older than 60 days Amount mismatch AI confidence below 80% Manager rejection Finance rejection Successful approval Document-processing failure Document the expected result, actual result, Pass/Fail status, and screenshots where useful. 16. Deliverables Power Platform Solution containing all required components Architecture diagram showing major components and data flow Dataverse design including tables, columns, relationships, and choices Final AI Builder Prompt and a brief explanation of its purpose Test document containing test cases, inputs, expected results, actual results, Pass/Fail status, and relevant screenshots README explaining architecture, flow responsibilities, Child Flows, Dataverse design, AI implementation, error handling, assumptions, and limitations 17. Completion Timeline You will have one week to complete the project. The one-week window is intentional so that the assignment can be completed alongside regular project responsibilities. Expected hands-on effort: approximately 14–20 hours depending on prior experience with Dataverse, AI Builder, and Power Automate. You are expected to design the solution yourself. You may use Microsoft documentation and other learning resources, but the final implementation should be your own. 18. Evaluation Focus The project will be evaluated on: Correctness and completeness of the requirements Solution architecture and separation of responsibilities Dataverse design and data relationships Effective use of Child Flows Quality and reliability of AI Prompt implementation Document-processing implementation Business-rule implementation Error handling and resilience Maintainability, naming, and use of Solutions/Connection References/Environment Variables Testing coverage Overall quality of the final solution

Project ID: 40674618