Ignite - AI for Mobile Monitoring
0 likes485 views
This document discusses using AI and machine learning to help with mobile app monitoring and testing. It notes that manual and scripted testing cannot keep up with the complexity, bugs from continuous integration, and deployment cadence of mobile apps. The proposed solution is to train an AI using examples of good and bad app behavior and performance on different screens to help with input validation, alerting, and prioritizing issues. This could generate tests and verifications without needing to know all failure modes ahead of time, saving both time and money compared to hand-crafted testing and maintenance.
Ignite - AI for Mobile Monitoring
- 1. AI for Mobile Monitoring Jason Arbon
- 3. Problem
- 4. Fix Attempt
- 5. Solution
- 7. Source of Production Bugs Continuous Integration (CI) Continuous Delivery (CD) Microservices/Operations App store Deployment Cadence ….Complexity
- 8. Manual Testing can’t Keep up
- 9. Scripted Automation can’t keep up
- 10. Log ‘GREP’ not enough (UX)
- 11. Welcome Machine Learning (aka AI) Robots just generated basic tests ...and generate verifications Repeat Mindlessly in production
- 12. Problems to Solve via AI Input Validation Alerting/Prioritizing
- 13. Input
- 14. Input
- 15. Training The AI Examples of Good Behavior ● Good (or Bad) Screens ● Good (or Bad) Performance for each screen
- 16. Training The AI
- 18. Verify Behavior
- 19. Benefits Don’t need to know all failure modes ahead of time! Code is ‘Generated’ vs hand-crafted (time + $) No code maintenance (time + $) Sounds clever ($) @appdiff: team is already monitoring 1000’s of apps