The Challenge
NTUC Digital needed to help employees navigate a complex landscape of benefits and services. With 33+ different services across multiple categories, employees often struggled to find the most relevant offerings for their specific situation.
Information was scattered across multiple sources, leading to confusion and high support inquiry volume. The traditional approach of static FAQ pages and search functions wasn't meeting the needs of a diverse workforce with varying career stages and life situations.
The Solution
We built a benefits navigation tool using AI-assisted development with Claude Code, Firecrawl for content scraping, and a lightweight RAG system for intelligent search.
AI-Assisted Development
Built with Claude Code for rapid iteration. AI-assisted coding let us ship faster while maintaining quality across the full stack.
Firecrawl Data Pipeline
Used Firecrawl to scrape and aggregate service information from NTUC sources, building a clean dataset of all 33+ services.
Lightweight RAG Search
Mini RAG system for intelligent search across services. Simple but effective retrieval matching user queries to relevant benefits.
Career Journey Matching
Quick assessment quiz matching employees to relevant services based on career stage and situation.
Development Process
The key to shipping quickly was combining AI-assisted development with tight feedback loops. Claude Code handled boilerplate and iteration speed, while we focused on product decisions and user testing.
Build-Test-Refine Cycles
Short development sprints with immediate feedback, allowing rapid course correction based on real user input rather than assumptions.
Quantitative Scoring
Measurable criteria for each feature, enabling objective assessment of progress and clear communication with stakeholders.
Integrated UAT
User acceptance testing woven throughout development, not just at the end, ensuring alignment with user needs at every stage.
Results
Successful UAT with high usability scores. The combination of Claude Code for development speed and Firecrawl for data aggregation let us deliver a working tool in a compressed timeline.
The project showed that AI-assisted development isn't about replacing human judgment—it's about spending more time on product decisions and less on boilerplate.
Testimonial coming soon
Want to Ship Faster with AI-Assisted Development?
I use Claude Code and modern tools like Firecrawl to build working software quickly. If you have a tool or internal app that needs to ship, let's talk.