AI is widely accepted in simple, transactional scenarios, where efficiency and convenience are valued. However, when it comes to complex financial decision making, a clear trust gap emerges.
While the industry continues to prioritise back office automation for cost reduction, AI is increasingly moving into front facing roles. This shift introduces new challenges, not only around regulatory and governance risks at an industry level, but more critically, around trust, transparency, and fairness from a user perspective.
Trust Breakdown in Financial Context
People do not reject AI outright in financial services, rather, trust weakens under specific decision making conditions. In finance, this breakdown becomes more likely when decisions are emotionally loaded, structurally complex, and tied to long term uncertainty.
Emotionally loaded
Financial decisions often carry immediate personal consequences affecting security, stability, and future plans. Under emotional pressure, users become significantly less willing to rely on automated recommendations and instead seek reassurance, control, and human validation.Structurally complex
Financial systems are inherently complex and often low transparency. When users cannot clearly understand how decisions are generated or what factors are being considered, opacity amplifies perceived risk, turning algorithmic efficiency into suspicion and reducing trust in AI supported outcomes.Linked to long term uncertainty
Many financial decisions involve long term commitments and uncertain outcomes, such as debt, investment, or relocation. As the time horizon extends and uncertainty increases, users become more risk sensitive and less willing to use AI to control systems.
These conditions are not experienced equally across users. As a result, trust breakdown is also unevenly distributed. People who more frequently face high stakes, uncertain, and complex financial situations are more likely to hesitate, question AI recommendations, and seek greater reassurance before acting.
If trust breakdown is unevenly distributed, who experiences it most? Women
Particularly in early and transitional adult life stages, often navigate layered financial realities shaped by income variability, caregiving expectations, disrupted career progression, and lower confidence in high risk financial environments.
Key Life Moments for Women
Women navigate financial decisions throughout life, but certain stages bring more complexity and high impact moments.
They navigate layered financial realities during different life moments, including:
Income variability and pay disparities
Career interruptions related to caregiving
Balancing multiple financial roles within households
Lower confidence in high risk investment environments, etc…
Young Postgraduate Women Moving Abroad
Within this broader pattern, young postgraduate women (25-39) moving abroad face an intensified version of these conditions. They must make financial decisions across unfamiliar systems while dealing with fluctuating exchange rates, new cost of living structures, and the long term uncertainty of studying and living independently in another country.
Persona
By the time Lily just moved to the UK
By the time Lily lived in the UK for a while
An International Student’s relationship with currencies usually shows up in:
Having savings in their Home Currency
Need for New Currency
To develop an effective Intervention, we had to understand the: Functional, Social & Emotional relationship our users had to the desire of maximising currency exchanges.
Traditionally, users will have to exchange their money through their desired channels, having to adhere to the exchange rates offered by their banks/financial institutions
We identified a gap where our users encounter challenges in effectively maximising the value of currency exchange opportunities.
Design Hypothesis
We believe that an AI powered platform combining educational resources, real time market insights, and emotional support features can empower young postgraduate women moving abroad to navigate currency exchanges effectively while reducing emotional stress and financial risk.
Through AI, we can bridge the gaps between the user’s needs and desires and the lack of knowledge and control over the foreign exchange market.
We aim to use our intervention to address users' pains, simplifying and easing their way of life.
Hypothesis—> AI Centric Features
Workshop Validation
To validate our hypothesis, we facilitated three workshops, testing the concept both as a standalone platform and as an integrated feature within an existing trusted service.
This allowed us to explore not only whether users would engage with the intervention, but more critically, under what conditions they would trust it.
While our primary focus was on women, including male participants enabled us to better understand how trust in financial technology differs across user groups.
Key Insights
Social validation
Participants, particularly women, consistently sought confirmation from trusted, often human, sources before making financial decisions. Trust was rarely formed in isolation.Personalisation
Tailoring information to individual goals and contexts significantly increased perceived relevance and strengthened trust in the intervention.Accountability
Users questioned who or what they were trusting, especially in scenarios involving risk. Clarity around responsibility and recourse was critical to building confidence.
Smart Exchange is an AI powered currency exchange feature embedded within banking apps, designed to increase adoption by giving users greater control, transparency, and trust.
Increased User Control
Set a transfer cap so you have control over how much is exchanged, even during automation.Increased User Customisation
Set a up a “cushion amount”.Increased Accountability
Features built into a trusted platform.
1
FEATURE: Controlled Automation
Controlled Automation allows users to benefit from automated currency exchange while retaining a clear sense of control over how and when transactions happen. Rather than fully delegating decisions to the system, users define the boundaries within which automation can operate.
Users configure key parameters that define how automation should behave, including their target exchange rate, transfer amount, and minimum or maximum limits. The system continuously monitors real time exchange rate movements in the background and evaluates them against these predefined conditions. When criteria are met, users can either be notified to make a final decision or allow the system to execute the transaction automatically within the limits they have set.
By shifting automation from “fully automatic” to “human in the loop,” the feature reduces the perceived risk of losing control, helping users take advantage of favourable rates without constant monitoring or anxiety.
2
FEATURE: SMART PREDICTIONS
Predictive Decision Support helps users navigate the inherent uncertainty of currency exchange by transforming complex financial data into clear, actionable guidance. Rather than requiring users to interpret volatile exchange rates themselves, the system provides forward insights that support more informed and timely decisions.
The system combines historical exchange rate trends, real time market data, and user specific financial context, such as spending patterns, upcoming expenses, or personal financial goals (i.e. tuition payments or rent deadlines). Using this combined dataset, it identifies patterns, forecasts potential rate movements, and highlights optimal exchange windows.
These insights are then translated into user facing recommendations, such as suggesting when to exchange, when to wait, or when a favourable threshold is approaching. This reduces the need for users to monitor markets continuously or rely on guesswork.
By shifting decision making from reactive to proactive, the feature reduces uncertainty and cognitive burden. Users no longer need to interpret complex financial signals on their own, which increases confidence in their decisions. This sense of informed control plays a critical role in building trust in both the system’s recommendations and its automated capabilities.
3
FEATURE: Personalised & Empathetic Communication
Personalised and Empathetic Communication ensures that financial information is delivered in a way that is both relevant and emotionally supportive. It recognises that financial decision making is not purely rational, but often shaped by stress, uncertainty, and personal circumstances.
The system generates tailored alerts and notifications based on each user’s behaviour, financial goals, and current context. These messages go beyond raw data or technical updates by incorporating a conversational, empathetic tone to guide users through decision moments.
For example, instead of simply stating a rate change, the system frames the information in relation to the user’s goals (“You’re closer to your savings target”) or timing (“This might be a good moment to act”), making the message more meaningful and easier to act upon.
By aligning communication with both user goals and emotional context, the feature increases engagement and reduces decision related stress. It makes interactions feel more human and less transactional, helping users feel supported rather than overwhelmed.
What I Learned
My adorable teammates: David Soh, Chen Bowen, Madeleine Mai, Sanyogita Nikam, Wang Liboyang (me), Devika Malik. 💗
Depth of Research Shapes Depth of Insight
One of the most important learnings from this project was that the depth of research directly influences the depth of the problem framing and solution designing.
The more we engaged with users, their behaviours, and their contexts, the clearer the underlying issues became. This reinforced the importance of investing time in research not just to validate ideas, but to uncover the right questions to design for.
The real challenge lies in simplifying the core tension
I realised that I was trying to capture too many layers at once without clearly defining a core tension. This made the problem feel complex but less focused.
It also made me question how much complexity is actually necessary, and where “simplification” could create more clarity without losing depth.
From Concept to Real World Constraints
This project highlighted the gap between a well developed concept and its feasibility in real world contexts. While the solution was conceptually strong, it required deeper consideration of implementation conditions such as integration with existing systems, institutional constraints, and stakeholder alignment.
This made me realise that strong ideas alone are not enough in some situations. Designing for real impact requires engaging with the practical conditions that shape whether a solution can exist beyond the concept stage.
This project represents the first time I have, as a service designer, attempted to address a real world problem.
Although the project was not ultimately implemented, the process provided valuable insights and learning opportunities. I remain deeply grateful to my tutor Nicolás Rebolledo Bustamante, my teammates, and the effort I invested throughout the process.