TrustRails

TrustRails is an AI-commerce infrastructure platform I built that standardises messy electronics retailer product feeds into one predictable API for AI shopping agents. It sits between retailers like AO.com and AI assistants like ChatGPT, Claude and Perplexity, normalising chaotic product data into clean, structured results with accurate live pricing and direct purchase links.

Project Purpose and Goal

AI shopping agents are becoming the new search engines but they struggle to recommend products because retailer data is inconsistent, scraped and unreliable. TrustRails transforms retailer feeds into one API and one schema with live structured data. The goal was to make it trivial for AI applications to search and surface real product recommendations with accurate, live pricing.

Webstack and explanation

The platform is built with Next.js and React with TypeScript throughout. The data ingestion pipeline integrates with affiliate networks including Awin, CJ Affiliate and Google Merchant feeds. I published two npm packages: a TypeScript SDK for developers to search and retrieve products, and an MCP server that lets AI assistants like Claude natively search the product catalogue as a tool during conversations.

Lessons Learned

This project taught me a huge amount about building developer tools and publishing npm packages. I learnt how to build an MCP server from scratch and integrate it with Claude and other AI assistants. Working with multiple affiliate network APIs and normalising inconsistent data formats into a single schema was a significant challenge.

Other Projects