Idea
A web app automating software issue detection and code revision for developers and QA teams to improve code quality efficiently
Research Paper
Core Innovation
This paper presents WALL, which uniquely combines SonarQube with large language models to automate the entire code quality assurance pipeline. Unlike prior tools that focus on detection or revision separately, WALL integrates issue extraction, automated revision, and evaluation in one platform. It also demonstrates cost-effective use of hybrid LLMs to balance quality and expense.
Market Size (TAM)
$2–10B TAM, $1–2B SAM; assumption: software development and QA automation market growth driven by increasing code complexity and demand for efficiency.
Potential Customers & Pain Points
- Software Development Teams Needing Faster Issue Detection
- QA Teams Seeking Automated Code Revision
- Enterprises Managing Large Codebases with High Issue Volume
Business Model
Subscription-based SaaS platform with tiered pricing based on codebase size and LLM usage; potential API access for integration.
Competitive Landscape
- SonarQube
- DeepCode (Snyk)
- Codacy
Implementation Challenges
- Dependence on LLM accuracy and updates
- Integration complexity with diverse codebases
- User trust in automated revisions
Validation Strategy
- Pilot deployment with mid-size software companies
- Collect user feedback on revision accuracy and efficiency
- Iterate to reduce human intervention and integrate open-source LLMs
Research Paper Overview
WALL: A Web Application for Automated Quality Assurance using Large Language Models
Summary
WALL is a web application integrating SonarQube and large language models like GPT-3.5 Turbo and GPT-4o to automate software issue detection, code revision, and evaluation. It features three modules: issue extraction, code issues reviser, and code comparison tool, enabling a seamless pipeline for quality assurance. Experiments on 563 files with over 7,599 issues show WALL reduces human effort while maintaining high-quality revisions. A hybrid approach using cost-effective and advanced LLMs lowers costs and improves revision rates. Future plans include integrating open-source LLMs and removing human intervention for fully automated code quality management.