Case Study
AI engineering powers retail growth: UST CodeCrafter rapidly accelerates Android app development
OUR CLIENT
This multinational retailer operates thousands of stores and a robust e-commerce business around the world, offering a wide range of clothing and home goods. With a reputation for high-quality products at competitive prices, the company generates more than $5 billion in annual revenue.
THE CHALLENGE
Delivering Android parity with iOS without costly, time-consuming code rewrites
The retailer recognized that its iOS-first approach limited reach to a growing Android user base. Company leaders wanted to rapidly expand app offerings to Android users while maintaining a consistent brand experience across platforms. The prospect of manually rewriting applications in Kotlin—the Android coding language—was costly, time-consuming, and risked divergent feature sets between iOS and Android apps.
Converting a complex Swift codebase into idiomatic Kotlin, while preserving app logic, patterns, and performance, posed significant challenges. Framework and structural differences between
Apple’s framework, UIKit, and Android’s toolkit, Jetpack/Compose, as well as dependency gaps, such as missing view models, analytics mocks, and network layers, created the potential for compilation failures and runtime instability.
To move forward, the retailer needed a predictable, low-risk modernization path that preserved existing investments while accelerating time-to-market.
THE TRANSFORMATION
AI-assisted code conversion with human-in-the-loop AI validation
Using UST CodeCrafter, an innovative AI-assisted modernization framework and process, UST quickly converted Swift code to Kotlin. The approach strikes a balance between AI-driven automation and targeted human-in-the-loop validation, ensuring rapid code conversion at scale and reliable quality. UST CodeCrafter combines a proprietary Abstract Syntax Tree (AST) code parsing engine and control flow graphs (CFGs) with chain-of-thought large language model (LLM) generative AI (GenAI) prompts to accelerate delivery.
The engagement began with a validation phase, converting two representative applications into Kotlin to demonstrate feasibility and ensure stakeholder confidence. UST then progressed to the Swift codebase by refactoring a selected module to Kotlin. AI handled most of the code conversion with 78% first-pass accuracy. This left just 22% of the work effort to UST engineers who refined the code to address cross-module dependencies, framework alignment, and runtime performance, delivering reliable production-ready Android code. The engagement also delivered:
- A scalable, governed code conversion framework—UST provided the retailer with structured AST/CFG rules and playbooks, engineering GenAI prompt patterns, and a repeatable development pipeline that preserved file paths and directory structures. This reduces costs and accelerates development cycles for additional modules.
- Human-in-the-loop validations—UST helped the retailer’s app development engineers understand best practices to review and curate the AI-driven transformations to mitigate risks and ensure code quality with speed and scale.
- A proven path to Android parity—The AI-driven CodeCrafter framework reuses business logic, standardizes Kotlin implementations to accelerate feature delivery, and aligns with Android best practices.
- Architectural integrity—By preserving project structure and coding standards across platforms, UST AI and engineering experts helped the retailer establish a proven path for future modernization efforts.
THE IMPACT
Faster code conversion with 78% accuracy paves the way for accelerated app delivery
UST CodeCrafter delivered on its promise to the retailer—a reliable, low-risk, GenAI-driven method for converting its iOS app to Android while ensuring a consistent brand experience across platforms. The retailer achieved these benefits:
- Rapid Android app delivery—The pilot project was completed in just 146 hours. This expedited pace, from set-up through optimization, including parity gates for unit/UI tests and static analysis, demonstrated the power of UST CodeCrafter to accelerate code conversion projects with peace of mind from built-in controls. Efficiency gains will increase with each iteration, further accelerating code conversion cycles.
- Enhanced productivity—With a 78% accuracy rate for converted code, this AI-driven precision limited the scope of manual rework to just 22% of the code conversion effort, boosting productivity, accelerating time-to-market, and freeing developers to focus on innovation.
- Proven scalability—Nearly 100,000 Swift tokens were converted to approximately 130,000 Kotlin tokens. This impressive conversion effort demonstrated the breadth and scale of UST CodeCrafter’s capabilities.
- An extensible framework for AI-driven code conversion projects—The retailer can use this same UST CodeCrafter method for other code-to-code migrations where syntax trees and mapping rules are available, for example Objective-C to Swift, Java to Kotlin, Kotlin to Jetpack Compose, AngularJS to React, and React to Vue. Reusable AST/CFG rule packs will accelerate subsequent code conversion projects, enabling the retailer to leverage economies of scale.
“UST helped us reach Android parity quickly without risking our iOS investment. CodeCrafter gave us the confidence to scale.”
— Director of Mobile Engineering, Global Retailer
Ready to accelerate your application development timelines? Click here to find out how UST CodeCrafter can transform your code conversion projects.
RESOURCES
https://www.ust.com/en/alpha-ai
https://www.ust.com/en/what-we-do/digital-transformation/data-analytics