Employee Reimbursement Automation System
Budget / Salary₹600–1,500
TypeFreelance project
LocationRemote
Posted1 hour ago
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
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
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.