MINE PERFUME LAB

THE CHALLENGE

Mine Perfume Lab sells Italian, hand-made niche perfume with a high fragrance concentration across three European markets: Greece, Cyprus, and Bulgaria. Operationally, that means three different languages, three price structures, and three separate logistics paths.

The main barrier to growth wasn't acquiring new visitors — it was the decision phase on the website. Niche perfumes are a considered purchase. Without a physical sample, a buyer can't decide on a photo or a dry list of ingredients. A conversation with an in-store advisor solves that in five minutes, and the standard chatbots available in online stores failed at it.

THE DIAGNOSTIC MOMENT: WHAT ARE WE SOLVING?

Standard Shopify bots can only answer pre-set FAQs. They can't understand the complex hierarchy of perfume notes, sort fragrances into families (fresh, woody, oriental), or compare products to outside brands the buyer already knows.

So we changed the approach completely. Instead of a static catalog, we tied the whole web experience to a guided conversation. We didn't add the AI as a side plugin — we set it up as the central strategic layer of the site. It actively guides the buyer through choosing, testing, and the final purchase.

WHAT WE BUILT

We built a custom AI system embedded directly into Mine Perfume Lab's existing Shopify infrastructure. It isn't an off-the-shelf app — it's a strategic layer that connects the product catalog, languages, markets, and decision model into one conversational experience.

The system is made up of six key elements.

Multi-market detection

The system automatically recognizes a visitor from Greece, Cyprus, or Bulgaria and serves them local pricing, delivery and return terms, and correct data tracking.

Trilingual layer

Data in English, Greek, and Bulgarian lives in one database. The buyer can switch language mid-conversation.

4-section flow

The user chooses from four clear directions: help picking a fragrance, exploring sample sets, getting to know the brand, or help with existing orders.

Preference-based recommendations

The system doesn't ask for exact perfume names, which drives buyers away. Instead it guides the conversation through questions about favorite notes, mood, and context of use.

Sample-to-bottle

A built-in system with a 20 % voucher and an automated flow that takes the buyer from a sample set to ordering the full bottle.

Web search integration

In real time, the system finds outside information on the fragrances a buyer mentions as references and compares them to the specs in Mine's catalog.

Technical overview

Claude Code · v0 · Supabase · pgvector · Vercel · Shopify Markets · Telegram API

Code ownership

The system is built with no lock-in. It stays fully owned by Mine and keeps running smoothly, even if the collaboration ends.

SERVICES

PIKA. AI Lab · Web Architecture · E-mail Pipeline · Acquisition (Meta)

Do you sell products where buyers take their time and need advice and guidance?

An online store shouldn't be a static catalog — it should be a system that actively guides the buyer to the right decision.