All work

06 / Independent project

Can Buy Lah

An AI-assisted product-understanding and price-comparison MVP for in-store shopping in Singapore.

S

Project background

Chinese-speaking shoppers in Singapore often switch between translation, search, marketplaces and social platforms to understand unfamiliar packaging, prices and reviews.

T

Goal & role

I independently moved from user-problem mapping and product flow to AI output rules, error-case review and an accessible MVP.

A

What I did

I rebuilt the journey as capture → identify → standardise → search → understand → compare → decide. For failures involving size, flavour, discounts and vouchers, I created exact-match, near-match and material-difference rules, then adjusted price priority and information hierarchy.

01 Capture product and shelf price in context.
02 Organise product information, feedback and guidance.
03 Compare in-store and online references for a clear decision.
R

Result & review

Current CV metrics: 90%+ product-image recognition accuracy, 85%+ effective price-match success and 30 invited test users. The MVP has been tested across beauty, pharmacy, food and daily necessities, with rules iterated through error-case review.

90%+image-recognition accuracy
85%+effective price-match success
30invited test users