Curiella

The Problem

Discovering the products users love and knowing when to buy were separate jobs.

Shoppers had to search across stores, save the pieces they liked and keep checking prices. Finding the look was one job. Getting a good price was another. Bringing together over a thousand retail catalogs with disparate metadata made the work harder. Stores use different names and categories for the same kinds of goods. Details were missing or buried in unstructured text. Curiella needed one clear way to describe and find products.

Our Solution

We designed and built a platform that harvested over a million products from more than 1,000 retail stores into a single catalog. We used the latest AI technology to build customized recommendations around each shopper's taste based on usage and inputs, then added features like saved lists, price histories and price-drop alerts to enhance it further.

We built search and data systems to handle the differences between stores giving products a shared taxonomy—a common set of categories—and configured custom AI agents to sort messy retailer data into categories shoppers could understand. Vector search helps find products by meaning, not just matching words. Automated collection, classification and indexing handle routine work. People handle the exceptions.

Built for Lean Operations

• AI agents that sort the catalog. We designed classification agents with clear categories and rules. They validate results and reuse accepted mappings. Rules handle known cases; AI significantly improves ambiguous data.
• Search that understands the request. We combined massive vector indexes, keyword search and ranked results. We created recommendation rules that use likes, dislikes, clicks, views, and saved products to reflect individual taste.
• Price alerts with controls. We built automated agents that track price history, account for currency, respect each list's alert threshold and suppress repeat alerts.
• Automation to reduce maintenance. Scheduled jobs collect and process products. Health check agents regularly detect stalled work and start recovery. Indexing resumes from saved checkpoints rather than starting over. And automated blog agents, regularly research trends in fashion, write engaging blog posts and find supporting products within the catalog to publish fresh content monthly.

The Value

Shoppers can find what they want, save it and know when the price changes. Curiella has a useful reason to bring them back. Its data systems are built to keep the catalog moving without someone tending every job.