Get4x Redesign with GenAI

Project Vision

4xLabs is a Singapore-based fintech company specializing in digital solutions for the foreign exchange (forex) industry, with a mission to bring greater transparency and efficiency to the physical cash exchange market. It operates two primary platforms: Biz4x (a SaaS operating system for money services businesses), and Get4x (a consumer-facing currency exchange aggregator). The Get4x mobile app and website have long faced challenges in usability, consistency, and branding. This project therefore takes the opportunity to redesign Get4x — primarily focusing on the mobile app — from visual style to core functionality, aligning it with current industry trends. The goal is to better bridge modern travelers and traditional money changers while enhancing transparency and efficiency in the currency exchange experience.

Challenges

	Gain domain knowledge and understand regulatory requirements in the currency exchange industry, as well as emerging technologies such as GenAI design, to clarify opportunities, constraints, and current design norms.
	Revise existing interaction flows while leveraging new technologies to support efficient task completion and align with modern usage habits.
	Refine products' visual style to align with company branding and current industry trends, building user trust and reducing cognitive load.
	Design new interaction patterns that enable efficient use and align with industry trends, ensuring external relevance and meeting user expectations.

Current Issues & Project Scope

Identify target user goals, pain points, and define project scope.

In Southeast Asia, Oceania, Central Asia, and Africa, travelers in many countries still rely heavily on cash to cover daily travel expenses. As a result, they often exchange currency at local money changers (foreign exchange providers), which typically offer more favorable rates than banks. However, traditional money changers have relied on foot traffic and physical signage, with limited ways to promote their competitive rates digitally or reach inbound travelers. From travelers' perspective, accessing information about nearby money changers and comparing rates has also been challenging. To address this gap, Get4x was developed by 4xLabs, a Singapore-based fintech company, to bridge travelers and physical money changers. Through Get4x app and website, travelers can search for and compare real-time exchange rates offered by nearby money changers, and reserve a rate before visiting the store.

During my internship at 4xLabs several years ago, I primarily worked on new feature design for Biz4x, a Point of Sale (POS) and Financial Transaction Monitoring System (FTMS), and assisted in clarifying issues within Get4x. At the time, I was not deeply involved in its product design. Years later, I observed that many of these issues — particularly around usability, consistency, and branding — remain largely unresolved. I therefore saw an opportunity to redesign Get4x in alignment with the company’s brand and current industry trends, with the goal of better connecting modern travelers and traditional money changers while improving transparency and efficiency in the currency exchange experience.

I evaluated the Get4x app again with the current design and identified seven usability issues across three heuristic categories (according to Jakob Nielsen’s heuristic principles):

	Consistency and Standards (Users should not have to wonder whether different words, situations, or actions mean the same thing. Follow platform and industry conventions.): • Color usage violates common UI conventions and may confuse users: - Gray is used for focus and highlight states, even though it is typically associated with disabled or inactive elements. - Bright red is used for money changer rate tags, a color commonly reserved for error states or warnings, which may create unnecessary alarm or misinterpretation. • The visual style does not align with the company’s established brand colors, reducing brand recognition and trust. This inconsistency is also observed across the company’s other products. • The interface reflects UI design patterns from over a decade ago. Given the evolution of modern UI standards and user expectations, the design now feels outdated and may reduce perceived usability and trust. • The app adopts Material Design navigation patterns, which is not aligned with iOS platform conventions. This inconsistency may confuse users and violates established platform standards.
	Flexibility and Efficiency of Use (Design for both novice and expert users, allowing experienced users to perform tasks more quickly without overwhelming beginners.): • The current interaction flow does not take full advantage of larger screen sizes on modern devices. Updating the layout to better utilize available space, simplify interactions, and reduce the number of required taps would improve efficiency and support faster task completion. • Communication between customers and money changers is fragmented and lacks flexibility. Integrating messaging and coordination into a single platform would create a smoother, more intuitive experience for both parties.
	Aesthetic and Minimalist Design (Interfaces should not contain information that is irrelevant or rarely needed.): • The browse view displays excessive information (exchange rates) by default, leading to information overload. With limited filtering and prioritization options, users may feel overwhelmed and unsure how to begin.

After clarifying the key issues, I defined the scope of the project. Get4x offers both mobile app and website. The app is the primary service used by customers, while the website largely mirrors the app’s functionality, serving as a supporting channel and a way to promote the app to users. As a result, this project focused primarily on redesigning the mobile app, and took one or two screens of the website as an example to demonstrate how the new visual style and design patterns could be applied to the web experience. Given a planned timeline of three to four months, the scope centered on the core user journey and key features — exploration, reservation, and confirmation.

At this stage, in addition to addressing existing issues, I also developed initial ideas for improving current functionalities and potentially leveraging new technologies to enhance efficiency of use. I will share more details about my product thinking and design process in the following sections.

Design Strategies

Ensure efficient and effective execution.

To ensure the project is executed efficiently and effectively, and the new design meets standards of viability, desirability, and feasibility, I adopted the following methods throughout the process:

	Leveraging GenAI tools for research, ideation, and content refinement: I used Gemini to gather background information; experimented with vibe coding and ideation using MagicPattern, Figma Make, Lovable, AI Studio, etc.; and relied on ChatGPT to refine and polish textual content within designs.
	Conducting secondary research to build domain knowledge: By combining notes from my previous experience with online research, I developed a deeper understanding of user challenges, service objectives, market positioning, competitors, and regulatory constraints.

	Applying structured design frameworks: I used methods such as Jobs to Be Done (JTBD) and service blueprints to clarify target users, uncover unmet needs, and map the current service experience. I also leveraged interaction flows to visualize, refine, and validate user journey and interaction design ideas.
	Consulting subject matter experts (SMEs) throughout the process: To better understand the opportunities and constraints of emerging technologies, I consulted with two GenAI product designers and one research scientist specializing in GenAI, from early exploration through final design refinement.

User, Service, and Regulation Studies

Clarify customer types, workflows, and challenges while building domain knowledge.

During my internship at 4xLabs, I gained a foundational understanding of the company’s business model and customer base — serving both money changers and travelers, generating revenue primarily through Biz4x, and using Get4x as a traffic-acquisition channel. To deepen my understanding of user types and their challenges, I began with secondary research.

Through the research, I identified two primary user groups for Get4x:

	Travelers who need cash for travel expenses and prioritize either speed of exchange or the best rates to maximize their budget.
	Expats who live and work locally, often handling larger sums for personal or business needs and exchanging currency on a regular basis.

Since many countries in Southeast Asia, Oceania, Central Asia, and Africa still rely heavily on cash for small purchases, hawker stalls, and local transportation, travelers are often forced to exchange currency either upon arrival or right before departure. To do so, they need to discover nearby money changers, compare exchange rates and reliability, and visit physical storefronts to complete the transaction.

Common challenges travelers are facing include:

	Hidden costs when using credit cards, often due to unfavorable exchange rates set by merchants.
	Poor rates and high fees at airport and hotel exchange counters.
	Difficulty and time spent checking real-time rates across multiple money changers.
	Long queues during peak hours at licensed money changers that offer better rates.
	The risk of receiving counterfeit notes or incorrectly counted cash.

The Get4x app helps address many of these pain points by making it easier for users to explore and compare money changers, reserve a rate in advance, and guarantee that rate before visiting the storefront. The app also integrates navigation and audit-trail features into a single experience, resulting in a smoother end-to-end journey. Even though, there is still room for improvement as mentioned earlier. This project focuses on resolving the identified issues and expanding the functionality of core journey to support a more transparent, efficient, and user-friendly currency exchange experience.

In addition to user and service studies, I also clarified key terminology, such as the difference between the overview rate (mid-market rate) and the transactional rate (actual exchange rate). I further built domain knowledge in relevant regulatory and operational areas — using Singapore as a reference — including Know Your Customer (KYC), Anti-Money Laundering (AML), the Personal Data Protection Act (PDPA), Technology Risk Management (TRM) guidelines, reservation codes, and audit trails.

Design Exploration & Ideation

Define core intertaction flow, key features, information architecture, and visual style.

With a clear understanding of the workflow, I reimagined what an ideal interaction flow could look like.

Based on this flow, along with the identified issues and defined project scope, I began brainstorming potential solutions and key features that would help users complete tasks efficiently and achieve their goals. Key feature explorations included:

	Browsing and comparing nearby money changers for specific currency needs, with tools to narrow down results: • Map view: current location, actual exchange rates, money changer locations, contact information (call & message), distance, opening hours, available cash inventory, navigation, and filtering. • List view: actual rates, locations, contact information, distance, opening hours, available cash inventory, navigation, filtering, and sorting.
	Saving preferred money changers (and managing related notification settings): saved list.
	Asking currency exchange–related questions: integrated customer support and an AI assistant.
	Reserving exchange rates and reordering (login required): reservation form, confirmation, reservation records, and transaction history.
	Tracking current progress and updates: notifications with filtering.

To support a more efficient currency exchange experience and better align with modern travelers’ needs and usage habits, I explored how these functionalities could be redesigned and enhanced through a GenAI assistant. This approach enables more effective resource planning, streamlines the exchange process, and helps address gaps in customer support by providing timely, contextual assistance directly within the platform.

I also evaluated the pros and cons of traditional chatbots versus GenAI assistants, including a potential hybrid approach. Given GenAI’s strengths in dynamic problem solving, multilingual conversations, and contextual memory, it is better suited to currency exchange scenarios. Additionally, if the company plans to expand its service scope or support broader conversational needs in the future, a GenAI-based solution offers greater flexibility and scalability.

With the idea of leveraging GenAI, I conducted online research and consulted GenAI subject matter experts (SMEs), including one research scientist specializing in GenAI and two GenAI product designers. My references included GenAI design guidelines such as Shape of AI and the Horizon Design System, as well as YouTube talks, LinkedIn and Google Skills courses on GenAI and conversational design, and a range of consumer GenAI apps and products with embedded GenAI features. My goal was to follow research-based principles and current industry norms to ensure the interaction feels smooth, intuitive, and trustworthy for users.

I also visualized the GenAI interaction flow and iterated on it through ongoing discussions with SMEs, refining the concept into a more usable, feasible, and viable solution.

With the key features defined, my next step was to explore the information architecture.

In my initial concept, the Browse feature retained the original design, allowing users to switch between map view and list view, while introducing new capabilities such as filtering, sorting, and sharing. The AI assistant was designed as an independent feature, enabling users to complete exploration and reservation tasks through conversational interactions. I also placed the Saved List and Messages under the Account section, treating it as a hub for user-related information and settings.

However, after further consideration and consultation with SMEs, I revised this structure. I embedded the AI assistant into the browsing experience to streamline exploration and reduce context switching, and renamed the category to Explore to better reflect its expanded functionality. Given the available space and the needs of frequent users (particularly expats), I moved the Saved List to the first navigation level for quicker access. Since Messages and Notifications both represent incoming information that users need to review, I grouped them under a unified Inbox category. As the feature set matured, I also added Reservation Draft management to the Account section to support incomplete or ongoing tasks.

Moving forward, I used GenAI tools to support ideation and brainstorming. My vibe-coding process began with a brief concept description, followed by a more detailed prompt outlining key functionalities. I then iterated multiple times to explore different variations, and imported outputs into Figma when possible.

In parallel, I reviewed over 20 consumer apps, including those focused on online ordering, shopping, and trip planning experiences. Among them, more than 15 incorporated GenAI-related features, which helped inform patterns, expectations, and best practices for the design.

Before moving into visualizing specific design ideas, I began with branding and visual exploration. The original design suffered from several consistency issues: its color usage violated common UI conventions, the visual style did not align with the company’s established brand colors, and the interface felt dated.

My goal was to align the product’s visual language with both the company’s branding and current industry trends, creating stronger internal consistency while ensuring external relevance. This alignment helps build user trust through a sense of reliability, reduces cognitive load, and better meets user expectations.

To achieve this, I selected a dark theme as the foundation of the design. This choice respects 4xLabs’ branding, improves visual cohesion and recognizability across products, and aligns with contemporary design trends. The dark theme also conveys a modern, professional tone and helps users focus on core content. Building on this foundation, I refined the logotypes across 4xLabs’ products and used GenAI tools to explore and define secondary and tertiary colors, along with functional, neutral, and semantic color palettes.

In parallel, I worked with AI tools to identify typefaces that convey professionalism, trustworthiness, and elegance while maintaining high readability and alignment with current tech product design standards. I ultimately selected Inter for all body text and small UI elements. As an open-source typeface optimized for digital interfaces, Inter ensures consistent readability across devices and platforms, and reflects the clean, legible, and subtly elegant aesthetic commonly seen in modern fintech products.

Design Iteration & Improvement - Money Changer Browsing & Filtering

Streamline map view exploration.

The existing browse experience helps users discover money changers within a city, but it does not sufficiently support filtering based on specific needs — such as finding the lowest rates, highest reviews, or nearest options. The location and currency selection flow can also be simplified.

I assumed that users would prefer a map-first experience, so instead of requiring upfront location and currency selection, I redesigned the flow to land users directly on the map when they open the app. Users can then adjust the location and currency selection as needed, based on their specific goals.

To improve transparency and user control, I replaced the opaque “Recommendations” feature with a criteria-based filtering system that breaks decision factors down clearly. Based on early exploration, I identified exchange rate, cash inventory, reviews, distance, and operating status as the most important criteria. I initially explored distance and area-based filters, but later removed them, as there are relatively few money changers within a given area, and users can easily assess proximity through pinch-to-zoom on the map. Physical distance or shared area does not always translate to easier access either. I also considered retaining the “Recommendations” option, but ultimately removed it to avoid conflicts with manual filters. The final design focuses on four key criteria with "available inventory" and "review at lest 4.0" applied by default as recommendations to help users quickly narrow their results.

Besides, for iOS, I followed platform-specific navigation patterns, interaction behaviors, and the latest visual design guidelines to ensure consistency with current conventions and achieve strong external relevancev.

For the map view, I designed for three primary scenarios. When users are in a supported city, the app displays the corresponding city map with their current location and nearby money changers. The default currency selection reflects the most frequently exchanged currencies in that city, and users can tap a rate label to view a money changer’s profile. When users switch to a city they are not currently in, the map is divided into regions, highlighting the best exchange rates for the default currencies in each region — based on the assumption that exchange rate is users’ top priority. Users can tap a rate label to zoom into a region and explore all money changers within it. If users are not located in a supported city, the app defaults to the Singapore map view, where 4xLabs is based.

Design Iteration & Improvement - AI Assistant Interaction

Introduce intelligent exploration pathway.

Beyond the existing map and list views, I explored a new GenAI-powered discovery flow to unlock an alternative exploration experience.

My initial concept was to position the AI assistant as a standalone category. After further consideration and consultations with SMEs, I decided to embed the AI assistant into the browsing experience to streamline exploration, and renamed the category to Explore to better reflect its expanded functionality. I replaced the list view with the embedded AI assistant, as its use cases are largely covered by GenAI capabilities and it is used less frequently than the map view and AI assistant, based on my assumption. To better support expats who revisit the same money changers regularly, I moved the "Saved List" to the first navigation level, taking advantage of the freed-up space.

I also reworked the conversation structure — shifting from a single conversation thread to scenario-based channels driven by user filters and entry points. This evolved further into starting a new conversation by default, while allowing users to revisit previous ones. This approach helps reduce hallucination risks and prevents outdated information from misleading users.

Finally, the AI assistant evolved from offering general help to delivering a more guided and personalized onboarding experience through suggested queries. Along the way, I finalized the service logic: the AI assistant is available only after login to protect user privacy, ensure data security, and allocate system resources to engaged users more efficiently.

For the query input, I designed the initial experience with suggested prompts to avoid a blank starting state and help users get started quickly. Input flexibility is supported through typing, voice input, and media attachments. To keep the interface clean, concise, and intuitive, I grouped related functions together and aligned the interaction patterns with familiar GenAI tools. During iteration, I replaced the expand input control with a clear-all action. Since the input field automatically expands as users type —and queries in the currency exchange context are typically short — the expand option added limited value. This change also improved space efficiency on mobile design.

For the generation output, I initially structured the response to include a summary, a data grid with map context, follow-up query suggestions, additional actions (feedback, copy, voice playback, regeneration, and sharing), and data sources to ensure transparency and verification. After consulting with SMEs, I refined the visual hierarchy and introduced status tags to improve scannability and readability. I also made the data grid and follow-up suggestions more interactive, enabling users to review details and take next steps more efficiently.

Upon further consideration, instead of enabling reservations directly from the data grid, I guided users to the money changer’s main page to confirm real-time rates and store status before proceeding. This avoids potential misinformation, as exchange rates and operating hours may change while users navigate the app. To further build trust, I added a brief explanation of how the results were generated, bridging the generated content and follow-up suggestions. Finally, to support one-handed use and consider user finger dexterity as well as the scenario of multiple follow-up suggestions, I used recommended query buttons, making it easier for users to scan, tap, and continue the conversation.

Regarding the assistant’s scope, my initial idea was to expand beyond currency exchange to include broader finance-related questions — such as resource allocation guidance, investigation suggestions, and definitions or regulatory lookups — to increase user engagement. I also considered allowing open-ended conversations, as long as the topics were not risky or harmful, with the assistant responding to finance-unrelated questions before gently redirecting users back to finance topics.

However, this approach risked inefficient use of company resources and the collection of data that offered limited value to 4xLabs, which is focused on financial services. Taking the company’s strategic direction into account, I ultimately chose a more constrained approach: clearly and politely communicating the assistant’s scope, minimizing user frustration, and directly guiding users toward finance-related questions.

As the core user flow evolved and AI responses were refined, I also addressed other common scenarios beyond non-financial queries. The key design principles guiding these decisions were clarity and brevity, proactive interaction guidance, and dynamic error handling through clarifying questions.

Design Iteration & Improvement - Notifications & Reservation Tracking

Enable efficient reservation workflow.

Once users select a money changer, they can initiate the reservation process by completing and submitting a reservation form to lock in the exchange rate and amount. Upon submission, users receive a notification for status update and a confirmation that can be presented at the storefront. This project also covers the ideal end-to-end reservation flow to ensure a smooth and reliable experience.

Reservation is a login-required feature that relies on user background information. Through iteration, I introduced several key improvements: the pick-up time window is clearly displayed on both the order details and review pages; the actual transaction rate and calculated exchange amount are shown upfront to reduce cognitive load; and the form supports auto-saving when users exit.

Reservation drafts are stored under the user account with refined data-handling logic. Once saved, sensitive information (e.g., ID numbers) is masked to comply with the Personal Data Protection Act (PDPA). Accessing a draft requires re-authentication (such as a PIN sent via SMS), and drafts are automatically deleted after 14 days of inactivity.

Notifications proactively keep users informed, send timely reminders, and surface promotional updates to drive engagement. Through iteration, I made two key improvements: replacing generic notification icons with money changer logos — prioritizing the user’s relationship to a specific money changer over notification type — and shifting from a separate notification detail page to a tap-through model that takes users directly to the relevant item, supported by clear action buttons.

I also defined a comprehensive notification taxonomy, balancing business goals with user experience. The system supports eight notification types: reservation confirmation, reservation cancellation, exchange completion, feedback reminders, reservation draft reminders, reservation deletion reminders, rate-drop alerts (from saved money changers and frequently exchanged currencies, configurable in settings), and new money changer promotions — designed to benefit both users and money changers without becoming intrusive.

In addition, I designed flexible notification filters, including notification type, time range, location, money changer, and currency pair (exchange from/to), enabling users to quickly find and manage relevant updates.

Final Design

Demonstrate the core user flow – exploration, reservation, and confirmation.

Below are the hi-fi prototypes demonstrating the core user flow of the redesigned Get4x — from exploration and reservation to confirmation. In addition to mobile app (iOS) version, I also created a web mockup using the Explore first layer as an example to show how the new visual style and design patterns translate seamlessly to the web experience.

The core user flow begins with a map-based exploration experience, where users select their location (country and city) and desired currency pairs. When users are in the selected city, the map displays their current location along with nearby money changers. Users can pinch to zoom to adjust the distance scope and tap rate labels to view detailed money changer profiles. From there, they can contact the money changer, consult the AI assistant, navigate to the storefront, or reserve currency instantly, as well as save or share the listing. Beyond traditional map-based exploration, users can also take a more intelligent, conversational path by interacting directly with the AI assistant.

If users are not in their selected city, the map divides the city into regions and highlights the best exchange rate in each area. Users can tap a rate label to explore all money changers within that region. As users dive deeper, the bottom sheet updates accordingly and can be swiped up to reveal detailed information.

Users can switch cities and select the currency types they want to explore, then apply filters to narrow the results. By default, “Available inventory ≥ 1,000” and “Review rating ≥ 4.0” are preselected as recommended filters to help users quickly find reliable options. Users can further adjust these filters to refine the results based on their preferences.

Users can also explore using the AI assistant by starting with suggested prompts or through text, voice, or media input. Based on the assistant’s responses, users can decide to proceed with the presented options or continue the conversation to further refine their choices.

Users can make a reservation directly from the money changer’s main page within the AI assistant. Information collected during registration is auto-filled to reduce effort, requiring users to complete only the remaining order details, customer information, and payment details. If users exit the form before submission, their progress is automatically saved as a draft for later completion.

Users can view their reservation confirmation and details directly within the AI assistant. After submitting a form, suggested next-step queries appear to help users seamlessly continue their journey. For any special requests or needs, users can also ask the AI for assistance.

Once a reservation is submitted, users receive a notification confirming its status. Tapping the notification takes them to the reservation detail page, where they can view both active and expired reservations, and easily make a new reservation if needed.

Users can access the same services on the web, designed to closely mirror the mobile experience for seamless cross-platform consistency. The AI assistant appears either as a side panel alongside the map view or as a compact floating interface above it.

Reflection

Limitation

	The project primarily addressed the core user flow.
	Additional edge cases and exception scenarios remain to be explored.

Next Step

	Validate assumptions about user behavior and preferences by testing key design decisions with real users (e.g., map-first exploration, contextual AI assistant entry points, etc.).
	Measure trust and experience quality through adoption, retention, task success, and user satisfaction metrics.
	Integrate human customer support with the AI assistant to provide seamless, end-to-end assistance.
	Introduce dark mode to reduce eye strain, improve accessibility and inclusivity, enhance focus, and optimize battery usage.

What I Learned

	Product design must account for regulatory constraints where compliance directly shapes user flows and feature feasibility.
	Consulting with experienced practitioners helped refine functionality and surface blind spots early.
	AI assistant interaction design deserves deeper exploration, as it directly impacts usability and user-perceived value.