Project Overview
Managing enterprise IT infrastructure for hundreds of daily workers involves handling a massive volume of repetitive first-level queries—ranging from printer connections and Wi-Fi configurations to password resets.
The Problem: Systems administration teams spent substantial work hours answering basic, repetitive support requests. This delayed resolving critical infrastructure challenges and backlogged service desk operations.
The Solution: I built a conversational AI IT Support Assistant integrated into the corporate intranet. The chatbot analyzes user problem descriptions, performs a semantic search on our technical knowledge database, and uses the Google Gemini API to write tailored, step-by-step resolution guides for the employee. If the issue remains unresolved, the bot automatically files a categorized ticket directly into the administration team's dashboard.
Key Capabilities
AI Assistance
Interprets user descriptions of technical problems and explains technical resolutions in simple, clear language.
Knowledge Base Search
Queries localized reference documents to find system configurations, network rules, and hardware troubleshooting guides.
Automated Ticket Creation
If first-line assistance fails, the assistant compiles a structured error report and files a support ticket automatically.
Department Escalation
Classifies incoming tickets by category (Hardware, Network, Accounts) and automatically routes them to the correct technician queue.
Lessons Learned & Insights
Deploying generative AI inside private networks requires careful prompt structuring and routing rules. We learned that the LLM must be tightly constrained to references from our internal technical manuals, preventing "hallucinations" regarding network configurations.
By routing queries through a local pre-filter (verifying keywords against known documentation topics), we optimized API costs and latency, providing users with instant answers for frequent queries.