Map Your App for AI
summary
Block’s open-source CLI Trailblaze turns LLM-driven app exploration into replayable YAML trails that run deterministically in CI and locally without hitting LLM APIs on replay. It records screenshots, full view hierarchies, Logcat, networking and analytics, generates a standalone zip artifact, and supports Android, iOS, web and desktop drivers. Teams define custom tools and trailhead start states; Block runs seven target families with hundreds of nightly trails, and recordings make failures triageable and reproducible.
AI-written summary · about
chapters (11)
- 0:00Introduction to Trailblaze
- 3:04What is Trailblaze
- 9:22Demo of Trailblaze reports
- 12:52Why Trailblaze uses recordings
- 18:05How Trailblaze creates reliable trails
- 22:54Understanding trail maps
- 28:45Visualizing app navigation
- 32:15Creating a trail map for DroidCon
- 33:38Custom tools in Trailblaze
- 39:14Type safety in Trailblaze
- 40:40Live demo and Q&A
related talks
41mAI Driven Mobile Testing
12mAnswering Developer Questions with Annotated Agent-Discovered Program Traces
15mAI in the Terminal: Empowering Developers with CLI-Based AI Apps
16mYou can’t write “AI” without an “I” Control and responsibility in the age of copilots
38mQuerying Live Data With LLM App
26mMobile-First Setup for AI Coding
46mBringing Mobile App Validation into the Agentic Loop
54mAI for Software Testing: The Ultimate Journey