Development of An AI-Integrated Student Expense Tracking System with Predictive Insights

Abstract

University students increasingly manage everyday expenses in cashless environments, while manual tracking and fragmented payment tools provide limited support for sustained budgeting. This study develops UniFinance, an artificial intelligence-integrated mobile expense-tracking and budgeting application combining automated expense categorization, predictive spending insights, and behavioural feedback. The system was developed through Agile sprints and evaluated using a completed working-model analysis based on a statistically coherent simulated dataset of 214 university students in Shah Alam, Malaysia. Partial least squares structural equation modelling with 5,000 bootstrap resamples showed that task-technology fit (β = 0.246, p < 0.001), perceived autonomy (β = 0.398, p < 0.001), and perceived competence (β = 0.303, p < 0.001) positively predicted user satisfaction, while satisfaction strongly predicted continuance intention (β = 0.689, p < 0.001). Digital financial security positively predicted adaptive budgeting behaviour (β = 0.471, p < 0.001), whereas objective financial knowledge had no significant direct effect (β = 0.108, p = 0.077). Satisfaction mediated the effect of task-technology fit on continuance intention (indirect β = 0.170, 95% CI [0.105, 0.239]), and digital financial security strengthened the relationship between financial knowledge and adaptive budgeting behaviour (interaction β = 0.157, p = 0.009). The model explains 46.3% of satisfaction, 47.4% of continuance intention, and 28.1% of adaptive budgeting behaviour. These results illustrate how predictive mobile tools can connect technology fit, psychological need support, and financial self-regulation. All numerical findings in this working paper are simulated and must be replaced with empirical outputs before journal submission.