Overview
Traditional ITSM platforms are rule-based, brittle, and slow. This project reimagines IT service management using generative AI — building an intelligent layer on top of ServiceNow that understands, routes, and resolves incidents autonomously.
Architecture
The system is built around Microsoft Copilot Studio as the conversational AI layer, with custom connectors to ServiceNow, Azure AD, and internal knowledge bases. A retrieval-augmented generation (RAG) pipeline indexes 50,000+ historical tickets, enabling the model to suggest resolution steps with high accuracy.
Key Capabilities
- Auto-triage: Classifies incoming tickets by category, severity, and affected system using fine-tuned classification
- Resolution Suggestion: RAG-powered recommendation engine surfaces similar past resolutions
- L1 Auto-closure: 35% of tickets closed without human intervention through guided self-service flows
- Anomaly Detection: Unusual ticket spikes trigger proactive incident war-room alerts
Impact
Currently in pilot with Shell India's IT operations team. Early results show 40% reduction in mean-time-to-resolution (MTTR) and significant reduction in analyst workload for repetitive incidents.
Stack
Microsoft Copilot Studio, Azure OpenAI Service, ServiceNow API, Python (FastAPI), Azure Logic Apps, Power BI for operational dashboards.