Agentic E-Care Portal

The Challenge

Just Dial's 15,000+ employees relied on 8+ disconnected internal systems for HR queries, IT support, leave management, payroll, and more. Each system had its own interface, login, and workflows — leading to high support ticket volumes, slow resolution times, and frustrated employees navigating multiple tools for simple tasks.

The goal was to unify these systems behind a single AI-driven conversational interface that could understand context, route queries intelligently, and take action across systems autonomously.

Key Features

🧠

Agentic AI Orchestration

Multi-agent system using LangGraph for complex, multi-step task resolution across modules with autonomous decision-making.

🔗

MCP Integration Layer

Model Context Protocol connects the AI to 8+ existing enterprise systems, enabling real-time data access and actions without rebuilding existing infrastructure.

💬

Conversational Interface

Natural language queries for leave balances, payroll details, IT tickets, policy lookups, and more — all from one chat window.

🔒

Role-Based Access

Integrated with existing IAM/SSO to ensure employees only see data they're authorized to access, with full audit trails.

📊

Analytics Dashboard

Real-time monitoring of query volumes, resolution rates, agent performance, and system health for operations teams.

🔄

Smart Escalation

AI agents detect when human intervention is needed and seamlessly escalate with full context to the right support team.

Technical Approach

Agent Architecture

Designed a multi-agent system with LangGraph where specialized agents handle different domains (HR, IT, Finance) and a router agent orchestrates conversations and delegates tasks.

Context Protocol

Implemented MCP to create a standardized interface between the AI layer and legacy enterprise systems, enabling tool-use patterns without modifying existing backends.

RAG + Knowledge Base

Built a RAG pipeline over company policies, SOPs, and documentation using vector embeddings, ensuring the AI provides accurate, sourced answers.

Outcomes

~80% auto-resolution rate — majority of employee queries resolved without human intervention
Single interface — replaced 8+ separate systems for common employee queries
15,000 employees — serving the entire workforce in real-time
Reduced support load — IT and HR support teams freed up to focus on complex issues
Faster resolution — average query resolution time dropped from hours to seconds