HRM Chatbot Integration

Overview

This repository contains the backend logic and query registry for the HRM Chatbot Integration. The chatbot allows users to query HRM attendance data using text or voice input in a safe, controlled, and enterprise-ready manner.

The design ensures zero hallucination risk, strict data access control, and predictable behavior suitable for HR and compliance-sensitive environments.

---

Scope (Phase 1)

The initial implementation supports attendance-related queries using the following tables:

• AttendanceDaily – Daily attendance and real-time status
• AttendanceSummary – Monthly and period-based summaries

Future phases may extend support to additional HR modules.

---

High-Level Flow

1. User opens the HRM application.
2. User interacts with the chatbot via:
   • Quick suggestion buttons
   • Text input
   • Voice input
3. The application sends the request to the backend chatbot API.
4. Input is converted to text (if voice).
5. An intent is detected using rule-based and AI-assisted matching.
6. The intent is mapped to a predefined SQL query.
7. The query is executed using read-only database access.
8. Results are formatted into a human-readable response.
9. The response is returned to the application as text (and optionally voice).

---

Safety & Data Protection

This chatbot is designed to protect HR data at all times.

• No dynamic SQL generation
• No free-form AI-generated answers
• Strict intent-to-query mapping
• Read-only database credentials
• Deterministic outputs for the same inputs
• No modification of HR data

AI is used only for:

• Intent detection
• Typo tolerance
• Natural language normalization

AI is not used to generate data or make assumptions.

---

Architecture

• Frontend: Flutter (HRM Application)
• Backend: .NET API (Chatbot Service)
• Database: SQL Server
• Access Level: Read-only
• Intent Engine: Rule-based + AI-assisted matching
• Query Execution: Parameterized SQL only

---

Repository Structure

Chatbot-Integration-for-SmartPruner/
│
├── README.md                     # HRM Chatbot integration overview & design
├── attendance_daily.md           # Intent + SQL mapping for AttendanceDaily
├── attendance_summary.md         # Intent + SQL mapping for AttendanceSummary
│
├── hrm-chatbot-test/             # Chatbot PoC / test harness
   ├── app.py                    # Local chatbot test runner (CLI-based)
   ├── intent.py                 # Intent detection logic
   ├── queries.py                # Intent-to-SQL mapping registry
   ├── db.py                     # Read-only database connection & executor
   └── __pycache__/              # Python cache (ignored in production)

Supported Features

• Daily attendance queries
• Monthly attendance summaries
• Employee-specific queries
• Count-based analytics
• Overtime tracking
• Late and absence tracking

---

Voice Support

Voice queries follow the same execution flow as text queries:

• Voice → Speech-to-text
• Text → Intent detection
• Intent → SQL query
• SQL result → Text response
• Optional text-to-speech response

---

Deployment Notes

• Chatbot logic is deployed as part of the backend API.
• Flutter consumes the API responses.
• No database schema changes are required.
• No additional tables are created.

---

Design Principles

• Safety first
• No hallucination
• Predictable outputs
• Audit-friendly
• Enterprise-ready
• Easy to extend

---

Author

Designed & Implemented by: Arif

---

License

Will be added in the Future :)