AI Unifies a 26M-User BNPL Platform Across iOS and Android

The Challenge

The Buy Now, Pay Later (BNPL) sector has become a defining force in consumer finance. According to Fintech Futures, the global BNPL market is projected to reach $560 billion in 2025, growing at a CAGR of 13.7%, driven by rising consumer demand for flexible payment options and deep integration across digital retail channels. As BNPL providers scale to serve tens of millions of users, their mobile applications have become mission-critical business infrastructure.

In this environment, mobile platform reliability, consistency, and the speed at which new features can be delivered are direct competitive differentiators. Many BNPL providers find it challenging to maintain a consistent, high-quality mobile experience across both iOS and Android risk losing users to competitors in an increasingly crowded market.

The client serves nearly 26 million users and processing $37 billion in annual payments. They had built and maintained two separate native mobile applications: one for iOS and one for Android. Over time, this dual-codebase model created compounding operational challenges. Engineering teams were required to implement every feature, fix, and update twice, once per platform, leading to growing maintenance overhead, platform drift, and the constant risk of inconsistent user experiences.

As the company's product surface expanded across retail, professional services, and financial management features, the cost and complexity of sustaining two independent native apps became a material constraint on the organization's ability to move quickly. The company launched an initiative to consolidate both native apps into a single React Native codebase, establishing a unified mobile platform capable of supporting the company's scale and growth trajectory.

In brief the client needed to:

• Eliminate duplicated engineering effort by replacing two native codebases with one
• Achieve full feature parity across iOS and Android from a single shared implementation
• Accelerate feature delivery by building once and shipping to both platforms simultaneously
• Establish production-grade operational standards covering observability, security, and accessibility
• Implement a controlled, phased rollout strategy to manage risk during the transition
• Modernize authentication to give users a consistent login experience across platforms
• Create a scalable mobile platform capable of supporting continued user and merchant growth

Our Solution

GSPANN contributed a focused engineering and program management team to the project, a large-scale initiative involving approximately 55 engineers across multiple workstreams. Our team, comprising React Native developers and a project manager, worked within the client's structured delivery model to migrate consumer-facing screens and features from the native iOS and Android codebases into the new shared React Native application.

Rather than rebuilding the entire product from scratch, the program adopted a pragmatic hybrid approach: complex, frequently changing flows (such as form- heavy financial journeys) were preserved by embedding existing web experiences inside native wrapper screens using the client's internal SDK. This avoided unnecessary rework while still delivering a native entry point and navigation experience.

New and migrated screens were built in React Native with TypeScript, following the client's evolving design system and component patterns. It's worth noting that the GSPANN development team contributed to the optimization of the design.

The delivery model was operationalized through a structured migration tracker that moved each feature surface through defined phases: ownership claim, effort sizing (SWAG), specification, implementation, and UAT readiness. This gave leadership real-time visibility into scope, dependencies, and readiness across a large multi-team program, replacing informal coordination with a shared, auditable source of truth.

AI-assisted development was embedded into the delivery process as a formal practice, not an ad hoc tool. The client ran a dedicated AI enablement initiative, and the migration tracker included an AI-related KPI flag to make AI's contribution to delivery measurable at the work-item level. AI tools that included Cursor and Claude Code were used to accelerate screen scaffolding, component generation, E2E test authoring, and React Native pattern translation from native code. A key enabler was the use of an MCP (Model Context Protocol) server to feed structured Figma design context directly into AI coding tools, significantly improving the accuracy of UI implementation and reducing iteration cycles.

Quality assurance was distributed across teams rather than centralized. Each product team owned the sign-off for its own Critical User Journeys (CUJs), validated against shared builds using a structured UAT process. End-to-end tests were authored using Maestro, with a target of full P0 CUJ coverage. Sentry and Chronosphere dashboards provided continuous observability on app health, latency, and success rates throughout the ramp. Static and dynamic analysis via NowSecure and penetration testing along with other security readiness aspects were treated as a launch prerequisite, not an afterthought.

Key success factors included:

• Executed a 9-month rollout to the first 1% of users and a 14-month full Android rollout, with iOS to follow
• Migrated 189 distinct screens and 191 screen registrations into a single React Native codebase
• Delivered 18+ feature modules and 5+ infrastructure modules across the consolidated platform
• Applied a hybrid implementation strategy, reusing existing web flows where appropriate
• Structured delivery using a phased migration tracker covering ownership, sizing, specification, and UAT
• Embedded AI-assisted development with tracked metrics on AI-contributed work items
• Leveraged Sentry and Chronosphere CUJ-centric observability dashboards and team-supported alerting to identify and fix bugs
• Used Maestro for E2E automation with a target of 80% unit test coverage across frontend flows

Lessons learned include:

• AI tooling requires upfront investment in enablement
• Distributed quality ownership improves actionability over centralized QA
• AI code generation effectiveness depends on iterative prompt engineering

Business Impact

• Unified code base: Single React Native codebase eliminates duplicate implementation across iOS and Android platforms
• Release on schedule: Android production launch achieved at the 1% to 5% and 50% to 100% rollout milestones well on schedule. The iOS production launch followed shortly after this.
• Crash-free rate above 99%: Crash-free rate of 99.89% confirmed at the 25% Android ramp go/no-go checkpoint, and a crash-free rate of 99.84% was maintained at approximately 5% rollout (~280,000 active users)
• Early adoption metrics: 13,000 users and 93,000 daily app opens within the first ramp window
• All tracked thresholds met: CUJ latency and success rates met go/no-go thresholds across all tracked critical user journeys
• Reduced cycle time: AI-assisted development measurably reduced cycle time for UI implementation and E2E test authoring
• Reduced rollout risk: Distributed CUJ ownership model reduced rollout risk by eliminating single-team quality bottlenecks
• Phased ramp with rollback: Phased feature flag rollout via Statsig enabled safe, incremental ramp with rollback capability
• Real-time delivery tracking: Structured migration tracker provided real-time delivery visibility across 55+ engineers and multiple teams
• High security standards: Production-grade security posture established through NowSecure analysis and penetration testing requirements
• Scaled rollout: Platform now positioned to support continued rollout to the client's full base of 26 million users