Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/yocxy2/2a/llms.txt

Use this file to discover all available pages before exploring further.

The AI Ticket Support System is a full-stack platform that uses artificial intelligence to triage, classify, and respond to customer support tickets automatically. Powered by GPT-4o-mini and a hybrid retrieval-augmented generation (RAG) pipeline, it reduces human agent workload by resolving high-confidence tickets without intervention.

Quickstart

Get the full stack running locally in under five minutes with Docker Compose.

Architecture

Understand how the backend, database, AI pipeline, and frontend fit together.

API Reference

Explore every REST endpoint — create, list, fetch, and update tickets.

Core Features

Learn how tickets move from submission to AI resolution or human review.

How it works

1

User submits a ticket

A customer describes their issue in the web UI. The React frontend posts the description to POST /api/v1/tickets.
2

AI classifies and responds

The backend generates a vector embedding, runs hybrid RAG search against the knowledge base, traverses the GraphRAG entity graph, then asks GPT-4o-mini for a category and response.
3

Confidence-based routing

Tickets with a confidence score ≥ 0.7 are marked ai_resolved automatically. Tickets below the threshold are set to pending_agent for human review.
4

Agents review and close

Support agents use the Admin Dashboard to inspect pending tickets, override responses, reassign categories, and close resolved cases.

Key capabilities

Hybrid RAG

80% vector similarity + 20% recency scoring + importance weighting for precise knowledge retrieval.

GraphRAG

Entity extraction and BFS graph traversal surface related concepts beyond simple keyword matching.

Async Processing

Entity extraction runs in background workers via Redis + BullMQ, keeping the API non-blocking.
You need a valid OpenAI API key and Docker installed before starting. See the Quickstart for step-by-step setup.

Build docs developers (and LLMs) love