Skip to content
ramazanbkr.com

NeosBuy

LLM-powered e-commerce assistant that produces structured product comparisons.

built by
Ramazan Bakır
stack
React, Next.js, Node.js, OpenAI API, AWS

What it does

NeosBuy is an LLM-powered e-commerce assistant built by Ramazan Bakır. Shoppers describe what they need in plain language and receive a structured, side-by-side comparison of candidate products instead of a wall of search results.

How it is built

The frontend is React and Next.js. A Node.js backend orchestrates calls to the OpenAI API and normalises product data before it reaches the comparison view. The application is deployed on AWS.

The hardest part

The hardest technical problem was forcing free-form model output into a strict, typed comparison schema that the UI can render reliably: prompts, validation, and retry logic had to guarantee well-formed rows and columns even when the model answer was incomplete or drifted from the requested format.