Kaura Natural / E-commerce / AI
An online health food store that understands what you're looking for.
Kaura Natural is an e-commerce store for natural products and specialized foods. I designed and built the platform from scratch, combining product design, user experience and artificial intelligence so that finding the right product is as simple as buying it.
- Product Design
- UX/UI
- Software Engineering
- AI Integration
The problem wasn't just selling online.
Shopping at a health food store often means needing guidance. Some people look for a specific product, but others have a need and don't know what to buy.
In physical stores, advice isn't always available. And in many e-commerce sites, finding something requires knowing its exact name.
The opportunity was to create a store that didn't just show products, but helped people discover them.
First, the information.
Before thinking about artificial intelligence, there was a fundamental condition: having good data.
I set a clear requirement for the client: every product had to be loaded with complete and accurate information. Not only name, price and photo, but also its characteristics, ingredients, properties and dietary restrictions, where applicable.
Information is the heart of the product.
01
Description
02
Ingredients
03
Categories and tags
04
Dietary information
This information doesn't just make for better product pages. It also feeds the search engine, improves result relevance and makes it possible to interpret more complex needs.
Without good data, not even the best AI model can deliver a good experience.
An experience built around how people search.
I designed the navigation, catalog, product pages and checkout around one idea: reduce the effort needed to find and choose.
But there was an opportunity to take that experience a step further. What happens when someone doesn't know a product's name, but can describe what they need?
Three ways to search. One experience.
A search that interprets, not just matches words.
Search by product
“Chocolate granola”
Search by brand
“Brand X”
Search by need
“I want something for breakfast, celiac-friendly”
How the intelligent search works
I proposed and built an intelligent search system able to combine products, brands, categories and needs expressed in natural language.
When someone searches for a celiac-friendly breakfast option, the system can identify the need, recognize the dietary restriction and find relevant products in the catalog.
AI helps interpret intent. The catalog data determines which products can be shown.
AI when it's needed. Engineering for everything else.
Not every query needs artificial intelligence.
That's why I designed a layered architecture that starts with traditional methods and uses AI models only when they add value.
Illustrative paths through the architecture, not a live execution or a guaranteed classification for these exact queries.
- Query
User search
Product, brand or need
- Preprocessing
Normalization
Text and entity cleanup
- Rules
Initial interpretation
Intent and confidence level
- Decision
Needs AI?
Confidence below 0.8
- Strategy
Select search
Based on the detected intent
- Textual
Full-text search
Keywords in PostgreSQL
- Results
Found products?
Check for matches
- Ranking
Sort products
Relevance, stock and sales
- Response
Show products
Real catalog results
- Caching
- Usage limits
- Fallback when AI is unavailable
The solution includes caching, usage limits and fallback mechanisms to keep search working when AI is unavailable.
The goal was to find a balance between result quality, speed, reliability and cost.
A product built to learn and evolve.
The platform includes admin tools to manage products, organize their information and analyze what users search for.
This makes it possible to see what people look for, what results they find and where there is room to improve.
It's not just about launching a store, but about building a foundation that can evolve with the business.
From idea to production.
I handled the whole process: understanding the problem, proposing the solution, designing the UX/UI, defining the architecture and building the platform from scratch.
Technologies
- Laravel 13
- Inertia.js
- Vue.js
- PostgreSQL
- pgvector
- Laravel Scout
- OpenAI
- Claude
- Jev
The result is an e-commerce that brings online sales, product information and intelligent search together in a single experience.
A store built so people can find what they need, even when they don't yet know exactly what to look for.