Jerome Nazario
← BackChapter II

Silangan Scents

Full Stack Developer · Scent meets AI · 2 min read

The Philippines has a thriving cottage industry of independent perfumers — small makers blending oud, vanilla, and citrus in batches of a few dozen bottles. Most of them sell exactly one way: face to face, bottle in hand. 3,000+ fragrance micro-businesses still run entirely on in-store sales, because the tooling to sell scent online simply doesn't exist at their size.

Silangan Scents is a full e-commerce SaaS built for that maker: product catalog, cart, checkout with the payment methods Filipino customers actually use, and one feature no template shop has — an AI Scent Finder that does the perfumer's counter-talk online.

Silangan Scents storefront — Eastern roots. Rare scents. Delivered.

A storefront that respects the product

Fragrance is a luxury purchase, so the store reads like one: editorial type, generous whitespace, photography-first product cards. Every scent gets its notes, its story, and its price — the same pitch the maker would give you across the counter.

Shop All — the full scent catalog

Checkout, the Filipino way

An online perfume shop in the Philippines lives or dies on its payment rails. Cards alone don't cut it, so checkout speaks all three local dialects:

GCash

The e-wallet in practically every Filipino pocket — the default way a customer under 40 pays for anything.

Maya

The second wallet, one tap away. Supporting both means never asking a customer to sign up for something new just to buy a bottle.

Cash on delivery

Still the trust anchor of Philippine e-commerce. For a first-time customer buying scent unsniffed, COD is what closes the sale.

The AI Scent Finder

The hardest part of buying perfume online is that you can't smell it. In the shop, the perfumer bridges that gap by asking what you like and handing you something close. The Scent Finder recreates that conversation: tell it a mood in your own words, pick a few notes you know you love, and the OpenAI API matches you against the actual catalog.

The counter-talk, coded

From “cozy rainy afternoon” to a bottle with a name.

Free-text mood and picked notes merge into one prompt; the model ranks real, in-stock scents rather than inventing poetry about ones that don't exist.

“cozy rainy”oud · vanillaOpenAI match1 · Shaghaf Oud2 · close match3 · wildcardmood + notesranked, in stock
Catalog-bound

The AI can only recommend scents that are actually in the store — no hallucinated products.

Two inputs

Mood is free text, notes are structured picks — loose enough to feel human, constrained enough to match well.

The AI Scent Finder interface

What comes next

The shop is live and the pattern is repeatable — catalog, local payment rails, and an AI clerk are exactly the kit those 3,000+ makers are missing. The SaaS shape means the next perfumer doesn't start from a blank page; they start from a store.

Tech stack
Visit live site