1. Executive Summary 

SKY HR is an AI-powered HR case management and workflow automation platform integrated with Microsoft Teams and a web-based HR portal. The solution enables employees to submit HR requests, ask policy-related questions, and receive support directly through Teams while allowing HR to manage cases, approvals, escalations, and employee communication through a centralized portal. The platform uses OpenAI-powered intelligence to classify requests, collect required information, generate summaries, and automate HR workflows. SKY HR improves response times, reduces manual HR effort, and provides a structured and auditable process for handling employee requests. 

Key Outcomes

Reduces manual HR request handling and triage effort.

Enables centralized case management and approval workflows.

Provides AI-powered employee self-service through Microsoft Teams.

Improves communication transparency between employees and HR.

Maintains complete audit trails for HR activities.

2. Problem Statement 

2.1 Background 

Organizations often manage employee requests through emails, chats, phone calls, and manual spreadsheets. This fragmented process creates delays, inconsistent responses, limited visibility into request status, and difficulty tracking approvals and resolutions 

2.2 Core Problem 

Employees lack a centralized and efficient mechanism to submit HR requests and obtain policy guidance, while HR teams lack a structured workflow for managing requests, approvals, escalations, and employee communications. This leads to delays, inconsistent processing, and limited visibility into request progress.  

2.3 Affected Stakeholders 

Stakeholder	Role	Impact	Key Concerns
Employees	Submit HR Requests	High	Quick Responses, visibility
HR Team	Case Processing	High	Efficiency, workload
Managers	Approvals, Escalations	Medium	Timely decisions
Leadership	Strategic oversight	Medium	Cost, ROI, delivery

2.4 Pain Points 

• HR requests are scattered across emails and chats. 

• Employees repeatedly ask the same policy-related questions.  

• Manual request categorization increases workload. 

• Limited visibility into approval and escalation status. 

• Delays caused by unavailable HR personnel. 

3. Solution Overview 

3.1 Proposed Solution 

SKY HR provides a centralized HR support platform combining Microsoft Teams, artificial intelligence, and workflow automation. Employees interact with an AI-powered Teams bot to ask questions, submit requests, and receive updates. The AI classifies requests, gathers required information, generates request summaries, and creates cases automatically. HR manage cases, approvals, escalations, and employee communications through a dedicated web portal. 

3.2 Out of Scope 

• Payroll processing.  

• Recruitment and applicant tracking.  

• Performance management. 

• Employee compensation planning. 

4. Screenshots & Workflow 

This section illustrates the end-to-end user journey through the system. Each step maps to a key interaction point. Replace the placeholders below with actual screenshots from the application. 

4.1 Workflow Overview 

Step	Screen / Action	Description	UI Location
1	User Authentication	User logs in via SSO or credentials	Login page
2	Dashboard / Home	User lands on personalized dashboard	Main dashboard
3	Initiate Action	User submits form, uploads data, or triggers job	Input form / Upload
4	Processing	System processes request, model runs inference	Progress indicator
5	Results & Review	Output rendered; user reviews and acts	Results view
6	Export / Handoff	User exports results or triggers downstream action	Export panel

4.2 Screen-by-Screen Walkthrough 

Step 1 Authentication 

Users authenticate using Microsoft Teams or the SKY HR web portal. Portal users authenticate using their registered email address and password.  

Figure 1 Authentication screen 

Step 2 : Dashboard 

After authentication, users with HR role are presented with a dashboard. 

Figure 2 . Dashboard with key metrics and navigation 

Step 3 : Input & Submission 

Employees interact with the SKY HR Microsoft Teams bot to submit HR requests or ask policy-related questions. User can also submit a case by asking for a specific form. Once all required information has been gathered, the system creates an HR case. 

Figure 3 Case Form 

Step 4 : Processing 

Upon submission, the system builds the case and displays a success message. 

Figure 4 Case processing and AI response generation 

Step 5 : Results 

Created cases are available through the SKY HR web portal, where HR users can review request details, communicate with employees, escalate cases, approve or reject requests, and monitor case progress.  

Figure 5 Cases Screen showing pending cases by default 

 Step 6 : Resolution & Notifications 

Once a final decision has been made, the system updates the case status, records the decision in the audit trail, and automatically notifies the employee and relevant participants through both the web portal and Microsoft Teams. 

Figure 6 . Final case resolution 

Figure 6 . Final Notification Delivery 

5. High-Level Architecture 

The architecture prioritizes scalability, maintainability, security, and user accessibility. 

Figure 7 High-level system architecture (replace with actual diagram) 

5.1 Architectural Layers 

Layer	Technology	Responsibility
Client Layer	Microsoft Teams	Employee interaction
Web Portal	Next.js, React, TypeScript	HR management interface
API Layer	ASP.NET Core Web API	Business logic
AI Layer	OpenAI GPT Models, Whisper	Classification and summaries
Database Layer	SQL Server	Data storage
Real-Time Layer	SignalR	Live updates
Identity Layer	Microsoft Entra ID, JWT	Authentication
Notification Layer	Bot Framework	Teams notifications

5.2 Data Flow 

The table below traces a request from user input through to final output, identifying the component responsible at each step. 

#	From		To	Description
1	Employee	→	Teams Bot	User submits request
2	Teams Bot	→	API	Request sent to backend
3	API	→	OpenAI	Classification and analysis
4	OpenAI	→	API	Prepared payload dispatched for inference
5	API	→	Database	Case saved
6	Database	→	Portal	HR retrieves cases
7	Portal	→	Employee	Updates sent through Teams

5.3 Model Choice & Rationale 

Attribute	Value
Selected Model	GPT-4o-mini
Model Type	Conversational AI / Classification
Provider	OpenAI
Context Window	Large-context support
Fine-Tuned	No

Why this model? 

• Strong HR request classification capability.  

• Natural conversational responses.  

• Fast response times.  

• Cost-effective compared to larger models.  

• Easy integration through OpenAI APIs. 

5.4 Integrations 

The system integrates with the following external systems. Each integration is described by protocol, authentication method, and purpose. 

System	Protocol	Auth Method	Purpose
Microsoft Teams	Bot Framework	Entra ID	Employee communication
OpenAI	REST API	API Key	AI processing
SQL Server	ADO.NET	Connection String	Data storage
SignalR	WebSocket	JWT	Real-time updates
Microsoft Entra ID	OAuth 2.0	JWT	Identity management

5.5 Security Considerations 

Security Posture Summary

HTTPS enforced for all communication.

JWT authentication for portal users.

Microsoft Entra ID authentication for Teams users.

Role-based access control implemented.

Internal and employee conversations separated.

Audit history maintained for all case activities.

Authentication & Authorization 

• Microsoft Entra ID for Teams authentication.  

• JWT tokens for portal access.  

• Role-based permissions:  

• Employee  

• HR  

• Manager  

• Team Lead  

• Admin 

Data Protection 

• Sensitive data stored in SQL Server.  

• API access secured through authorization middleware.  

• Case history and approval actions logged. 

Infrastructure Security 

• Secure API endpoints.  

• Role-based access restrictions.  

• Input validation and request sanitization. 

6. Deployment & Infrastructure 

6.1 Environments 

Environment	Platform	Purpose	Access
Local Development	Visual Studio, SQL Server, Microsoft Teams Toolkit	Feature development, debugging, and testing	Developers only
Frontend Production	Vercel	Hosts the HR Web Portal for end users	Employees, HR, Managers, Team Leads, Admins
Backend Production	Microsoft Azure App Service	Hosts the ASP.NET Core Web API, Teams Bot, SignalR Hub, and business logic	Authorized application users
Database	Microsoft SQL Server	Stores users, cases, approvals, workflows, policies, notifications, and chat history	Backend services and database administrators

6.2 CI/CD Pipeline 

1. Developers implement and test new features locally using Visual Studio and SQL Server. 

2. The frontend application is built and deployed to Vercel, making the latest user interface available to end users. 

3. The backend API is published from Visual Studio to Microsoft Azure App Service, deploying all REST APIs, SignalR services, AI integrations, and Microsoft Teams bot functionality. 

4. Azure App Service configuration (connection strings, JWT settings, Microsoft Bot credentials, and environment variables such as API keys) is updated through the Azure Portal without modifying application code.  

5. The Microsoft Teams application package is updated and uploaded to the Microsoft Teams Admin Center, allowing users to access the latest bot functionality. 

6. After deployment, functionality is verified by testing: 

• HR Portal authentication 

• Case creation and management 

• Microsoft Teams bot conversations 

• Workflow processing 

• Notifications and approvals 

• Database connectivity 

7. Open Issues & Risks 

Risk	Severity	Likelihood	Mitigation
AI misclassification	Medium	Medium	Human review workflows
OpenAI service availability	Medium	Medium	Retry mechanisms
Teams API limitations	Low	Medium	Monitoring and fallbacks
Data quality issues	Medium	Medium	Validation checks
User adoption challenges	Medium	Medium	Training and onboarding

8. Appendix 

A. Glossary 

Term	Definition
Case	An HR request submitted by an employee.
Adaptive Card	Interactive Teams card used for notifications and responses.
Entra ID	Microsoft’s identity and access management platform.
RBAC	Role-Based Access Control.
SignalR	Real-time communication framework.

FAQ’s