How Leading Tech Giants Are Solving Cross‑Platform Challenges with KMP and AI

Event Overview

On December 13 a technical salon on AI‑led cross‑platform development gathered experts from Kuaishou, Alipay, Tencent, and JetBrains to discuss Kotlin Multiplatform (KMP) and AI‑driven cross‑platform strategies.

Key Technical Presentations

KMP Evolution and AI Opportunities (JetBrains)

Fan Shengyou described Kotlin’s language evolution, the expansion of the KMP ecosystem, and introduced the AI framework Koog, showing how AI can automate code generation, UI adaptation, and testing in cross‑platform projects.

Kuaishou “One‑Code‑Many‑Platforms” Front‑End Architecture

Zhang Qingji presented a large‑scale front‑end architecture that shares a single codebase across iOS, Android, HarmonyOS, and Web. Core components include a platform‑agnostic UI layer built with Kotlin/JS, a shared business‑logic module compiled to native binaries via Kotlin/Native, and a dynamic module‑loading mechanism to keep app size low. The approach reduces manual adaptation while preserving UI/UX consistency.

Tencent KuiklyUI KMP/CMP Engineering Practice

Luoxuan introduced KuiklyUI, an open‑source UI framework built on KMP/CMP. The repository (https://github.com/Tencent/KuiklyUI) gained over 3000 GitHub stars within five months. KuiklyUI combines Kotlin Multiplatform for shared UI logic, Compose‑Multiplatform for rendering, and a Gradle plugin that automates platform‑specific resource generation. Deployment pipelines integrate with CI/CD to produce iOS, Android, and Web artifacts. Challenges discussed include binary‑size optimization, platform‑specific performance tuning, and incremental compilation.

Alipay Billion‑User KMP Migration

Song Jingxuan detailed the migration of the “My” tab, a billion‑user entry point, from native triple‑platform code to KMP. Migration steps were: (1) extract common business logic into a shared Kotlin module, (2) replace platform‑specific UI with Compose‑Multiplatform components, (3) use Gradle’s kotlin-multiplatform plugin to generate iOS frameworks and Android AARs, (4) perform a staged rollout with feature flags. Outcomes included a 30 % reduction in codebase size, 15 % lower memory consumption, and latency under 100 ms after optimizing Kotlin/Native interop.

Kuaishou Live‑Streaming KMP Implementation

Guo Xiaoming explained the technical selection for real‑time interactive live‑streaming. The stack uses KMP for shared signaling and UI, Kotlin/Native for low‑latency video processing, and a custom bridge to WebRTC native SDKs. Performance tuning involved: (a) enabling Kotlin/Native’s -Xopt-in=kotlin.native.concurrent to reduce GC pauses, (b) profiling with Android Studio Profiler and Instruments to keep frame time under 16 ms, and (c) using incremental compilation to keep full‑team build times under five minutes.

Round‑Table Insights

Panelists agreed that cross‑platform development and AI are complementary: AI can assist in code generation, test‑case synthesis, and UI adaptation, while KMP provides a stable foundation for multi‑device consistency.

Takeaways

Adopt a shared‑code architecture early to avoid duplicated effort across iOS, Android, HarmonyOS, and Web.

Leverage Kotlin’s multiplatform plugins and Compose‑Multiplatform to unify UI and business logic.

Use AI‑assisted tools such as Koog for automated refactoring and UI adaptation.

Plan migrations in stages with feature‑flag gating and thorough performance benchmarking.

Open‑source frameworks like KuiklyUI can accelerate adoption and benefit from community‑driven improvements.