AI Engineer · intermediate
AI-Powered Self-Service IT Support Bot
An intelligent chatbot using RAG to answer IT questions from a knowledge base — with fallback ticket creation.
Project overview
Employers are looking for "AI-augmented" support staff. This project proves you can leverage AI to reduce ticket volume and automate repetitive tasks — a top priority for modern IT departments.
Real-world use case
A mid-sized company with 500 employees wants to deflect common "how-to" questions from their human support team to focus on complex technical issues.
Tools and technologies
- Botpress/Voiceflow
- OpenAI API
- Google Sheets
- Zapier
Skills you can demonstrate
Prompt Engineering, RAG, Workflow Automation
Build roadmap
- Write your knowledge base
20–30 Markdown articles on the common issues — VPN, password reset, printer setup, MFA — each with a one-line symptom and a numbered fix. Consistent structure makes retrieval work. - Pick your bot platform
Botpress (free, flexible) or Voiceflow (visual). Build a hello-world bot first and confirm you can publish it to a web widget before adding knowledge. - Ingest the knowledge base
Upload the articles as a knowledge source and configure the bot to search them before generating anything. Ask ten questions and check which article each answer came from. - Add the LLM as the fallback
When no article matches, route to the model with a system prompt that enforces IT-support tone and says when it is unsure. Log every fallback so you can write the missing article. - Build the ticket escalation flow
When the bot cannot help, collect name, device, and issue, and write a row to a Google Sheet through Zapier with a ticket number. Reply with that number. - Measure deflection rate
Run your test questions, count answered versus escalated, and improve the weakest articles until deflection passes 40%. Put the final number on your resume.
Ways to extend the project
Use an AI agent to "triage" ticket priority (Low/Medium/High) based on sentiment and urgency before it even reaches a human.
How to showcase your work
- GitHub: Knowledge base docs + README explaining the bot logic and integration.
- LinkedIn: Screen recording of the bot answering a complex IT question and creating a ticket.
- Resume: "Developed an AI-driven support bot that reduced simulated ticket volume by 40% through automated self-service resolution."