Migrating to Jetpack Compose: How AI Accelerated Our Journey at Caper
Article Summary
Instacart's Caper team turned a multi-quarter Fragment-to-Compose migration into a dramatically faster effort using AI coding assistants.
Matt Kranzler from Instacart shares how their team migrated 130+ Android screens powering smart shopping carts from Fragments to Jetpack Compose. They developed a four-phase strategy that let them leverage AI to accelerate what would have been months of tedious refactoring work.
Key Takeaways
- AI reduced Phase 2 migration time by 5-7x, saving 300-350 engineering hours
- Four-phase approach decoupled navigation from UI refactoring to reduce risk
- Evolved from markdown guides to structured AI skills for consistent migrations
- Treated AI instructions as code with 325+ line migration guide
- Ran migrations parallel to feature work without blocking product roadmap
AI coding assistants fundamentally changed the economics of technical debt, making previously impractical large-scale refactoring projects feasible and fast.
About This Article
Instacart's Caper team had over 100 features built with Fragments mixed alongside XML views and Compose code. This created technical debt that made it harder to maintain their smart cart Android app long-term.
They built a four-phase migration plan. First came implicit Fragment hosts, then type-safe Kotlin DSL navigation. Next they converted Fragments to Compose using AI workflows with 17 steps and progressive context disclosure.
The AI-assisted approach let teams migrate code in parallel without slowing down feature work. They could modernize their codebase systematically while keeping their product roadmap on track.