K-Assist: AI-Powered IT Support Platform

AI-powered IT support platform that reduces repetitive help desk tickets through intelligent self-service and seamless Jira escalation.

Comment

K-Assist — AI-Powered IT Support Platform

An AI-powered internal IT support platform designed to reduce repetitive support requests by providing employees with self-service assistance, intelligent knowledge retrieval, and streamlined ticket creation.

Overview

K-Assist was developed as an internal proof-of-concept during my IT Operations Co-op at Kubota Canada.

The objective was to reduce repetitive IT support requests by providing employees with an intelligent assistant capable of retrieving answers from internal documentation before escalating unresolved issues to the IT team.

Instead of immediately creating support tickets, employees can search the knowledge base through a conversational interface. If the issue cannot be resolved, the application guides users toward ticket creation, reducing manual effort for the IT department while improving response time.

The Problem

Internal IT teams receive hundreds of repetitive questions every month, including:

  • Password resets
  • VPN issues
  • Printer configuration
  • Microsoft 365 problems
  • Software installation
  • Company handbook questions

Many of these issues already have documented solutions but employees often don't know where to find them.

This creates unnecessary tickets that consume valuable IT resources.

Solution

K-Assist provides a conversational interface that allows employees to:

  • Search internal IT documentation
  • Receive relevant troubleshooting steps
  • Ask follow-up questions
  • Escalate unresolved issues through Jira
  • Reduce repetitive support requests

The long-term vision is to integrate multiple enterprise systems into a single AI-powered support assistant.

System Architecture

Employee
    │
    ▼
 Next.js Frontend
    │
    ▼
Node.js / TypeScript API
    │
 ┌──┴─────────────┐
 │                │
 ▼                ▼
MongoDB      AI Model
Knowledge      APIs
 Base
 │
 ▼
Relevant Solution
 │
 ▼
Employee

If unresolved
      │
      ▼
 Jira Ticket

Tech Stack

Frontend

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • shadcn/ui
  • Zustand

Backend

  • Node.js
  • TypeScript
  • REST APIs

Database

  • MongoDB Atlas

Integrations

  • Jira REST API
  • Google Gemini API (Prototype)
  • Snowflake (Internal data integration)

Features

Intelligent Knowledge Search

Employees can search the internal IT handbook using natural language instead of manually browsing documentation.

AI-Assisted Responses

The platform can integrate with LLMs to improve answer quality when documentation alone is insufficient.

Ticket Escalation

If no suitable solution is found, users can continue directly to Jira for issue creation.

Modern Chat Interface

A clean conversational interface makes technical documentation easier to consume.

Scalable Architecture

The backend was designed with modular APIs allowing additional enterprise integrations in the future.

My Contributions

I contributed to:

  • Full-stack application development
  • Backend API implementation
  • AI workflow integration
  • Knowledge retrieval logic
  • MongoDB integration
  • Frontend development using Next.js
  • REST API development
  • Testing and debugging
  • Deployment support
  • CI/CD collaboration

Technical Challenges

Retrieving Relevant Knowledge

The largest challenge was ensuring users received the correct documentation instead of overwhelming them with unrelated articles.

This required designing a structured knowledge retrieval workflow that balanced speed with relevance.

AI Integration

Large language models produce helpful responses but cannot always be trusted without enterprise context.

The system was designed to prioritize internal documentation while allowing AI assistance where appropriate.

Workflow Design

Instead of replacing the IT help desk, K-Assist complements existing workflows.

Employees attempt self-service first, reducing unnecessary tickets while preserving existing support processes.

Lessons Learned

Working on K-Assist taught me how enterprise software differs from academic projects.

Key takeaways include:

  • Designing software around business workflows rather than technology
  • Building maintainable REST APIs
  • Working with production-ready frontend architecture
  • Integrating external services safely
  • Collaborating across the software development lifecycle
  • Understanding the importance of monitoring, documentation, and scalability

Future Improvements

Potential future enhancements include:

  • Microsoft Entra ID authentication
  • Role-based access control
  • Audit logging
  • Semantic search using vector databases
  • ServiceNow integration
  • AI-powered ticket categorization
  • Conversation history
  • Usage analytics dashboard

Technologies

  • Next.js
  • React
  • TypeScript
  • Node.js
  • MongoDB Atlas
  • REST APIs
  • Tailwind CSS
  • Zustand
  • Jira API
  • Google Gemini API

What This Project Demonstrates

This project demonstrates experience with:

  • Full-stack software development
  • REST API design
  • AI-assisted workflows
  • Enterprise application architecture
  • Knowledge management systems
  • Modern React development
  • Backend service development
  • Database integration
  • External API integration
  • Production-oriented software engineering

Status

Prototype completed as an internal proof-of-concept.

Some enterprise integrations and security features remain intentionally omitted because they were outside the scope of the prototype.

Built With

Share this project:

Updates