DragonCrawl: AI-Driven Mobile Testing Framework Achieves 91.6% iOS Pass Rate
Hey! So there’s this new mobile testing framework called DragonCrawl that's using AI to improve on traditional End-to-End testing for apps. A group of researchers put it together, and you can find their findings on arXiv under ID: 2607.28750. Unlike older methods that focused on crash detection, DragonCrawl checks user flows with each code update to make sure nothing crucial breaks. It uses GPT-4o's advanced features and has impressive pass rates—91.6% for iOS and 92.2% for Android, based on over 1,000 automated tests. This has reduced onboarding time from up to 120 hours to less than 4 and saved about 27 years of developer maintenance. It was officially announced on July 26, 2025.
Key facts
- DragonCrawl is an AI-driven mobile testing system for continuous regression testing.
- It uses generative intent-based reasoning with large language models, specifically GPT-4o.
- Achieves 91.6% pass rate on iOS and 92.2% on Android.
- Runs 1,013 automated tests continuously in CI/CD pipelines.
- Reduces test onboarding time from 96-120 hours to under 4 hours.
- Saves an estimated 27 developer years in test maintenance effort.
- Validates specific user flows on every code change, blocking commits that break critical functionality.
- Paper available on arXiv with ID 2607.28750.
Entities
Institutions
- arXiv