How FastGPT Transforms Ticket Handling and Boosts Efficiency by 90%

Background

The KuJiaLe ticket system, built by the testing architecture team, is used for online issue handling, tracking, and analysis.

We first review the ticket‑handling workflow, identify the difficulties faced by both requesters and handlers, and then explore how FastGPT can address these challenges.

Requester pain points:

Too much information required when reporting issues.

Handler pain points:

Many FAQ and duplicate questions, low handling value.

Long investigation chains, scattered tools, time‑consuming.

Complex business logic leads to high handling cost.

After years of operation, no effective optimization methods have been found for these pain points.

FastGPT offers new problem‑solving ideas, potentially improving handling efficiency and reducing costs.

FastGPT Overview

FastGPT provides a knowledge‑base Q&A capability, allowing existing FAQs and troubleshooting documents to be leveraged as SOPs for ticket handling.

Answers include “citation parts,” ensuring that AI‑generated responses reference the original knowledge‑base content.

Proposed Approach

Based on the identified pain points, we suggest the following AI‑assisted actions for different roles.

Requester ticket creation: Use ChatGPT to answer FAQ‑type issues automatically; if unresolved, let ChatGPT format the ticket according to a preset template, reducing manual effort and ensuring clear, standardized submissions.

Investigation phase: Retrieve relevant troubleshooting documents and tools automatically, extract key information (e.g., product ID, user account) using ChatGPT’s language analysis, eliminating the need to search across multiple environments.

Conclusion and solution output: ChatGPT determines the needed conclusions, suggests which data to fetch and which tools or APIs to call. The ticket system then executes these actions, returning results to the handler.

Implementation

Overall Architecture

FastGPT is integrated with the ticket system. FastGPT identifies the problem and decides which execution actions the ticket system should perform. The ticket system carries out these actions, retrieves data, and presents the results to the handler, achieving end‑to‑end automated analysis and resolution.

Concrete Features

Embedding FastGPT Q&A into the ticket system

Operators can consult the embedded FastGPT before submitting a ticket, resolving most FAQ issues automatically. The Q&A is linked to KuJiaLe’s internal knowledge base for accuracy.

Combining FastGPT knowledge base with troubleshooting tools

FastGPT matches issue descriptions (e.g., rendering slowdown) to predefined scenarios and supplies the ticket system with the required investigation steps, such as querying rendering time.

The ticket system executes data‑fetching APIs based on database configuration and returns results to the front end.

The front end formats the data, assisting handlers in decision‑making.

Fallback strategy for manual data retrieval

If automatic scenario recognition fails, users can manually specify required data, further improving investigation efficiency.

Results

For tickets related to rendering failures, manual investigation previously took 30 minutes; with FastGPT assistance it now takes about 3 minutes—a 90 % efficiency gain.

Technical support and incident tickets have seen average handling time reduced by nearly three days.

Future Outlook

Enable AI to automatically create tickets after processing user queries, further reducing requester effort.

Enrich ticket content automatically to allow more comprehensive AI‑driven conclusions, minimizing manual investigation.

Promote the proven workflow across other business lines to continuously improve efficiency.