Low on code, high on miles: S’porean uses AI to create tool to maximise credit card rewards

KiasuMiles, built by Hosan Swee, reflects a growing trend: AI tools are enabling non-coders to develop functional applications. Swee, an engineer without coding skills, used Codex and ChatGPT Work to translate plain English prompts into code. This approach allowed him to create a structured knowledge base of Singaporean credit cards, including complex rules like minimum spending thresholds and monthly miles caps. The AI models scanned bank and merchant websites in minutes, a task that would take hours manually.

This development points to a shift in Singapore's startup landscape. The barrier to entry for building specific, data-intensive tools is lowering. Small, niche applications can now be developed quickly and cheaply, challenging traditional fintech players. The ability to parse complex, distributed information from 50 credit cards and 3,300 merchants efficiently gives KiasuMiles a competitive edge in a crowded market. This model could be replicated for other complex consumer data sets across Southeast Asia.

The thing to watch is how quickly similar AI-driven tools emerge in other consumer finance niches. If KiasuMiles gains significant user adoption, it will validate the low-code AI development model for complex financial data. The test for Singapore's larger banks is whether they can integrate similar AI capabilities to offer personalized recommendations, or if they will be outmaneuvered by agile, AI-first startups like KiasuMiles.