Introducing Catalogica

If you sell products from many suppliers, you know what arrives. One sends a clean XML feed. One sends a spreadsheet with its own column names. One sends a price list as a PDF, and one sends a barcode and a name that only its warehouse understands. Every one of them expects your team to turn that into a product page a shopper can buy from.

Catalogica does that work. It takes whatever your suppliers send, from a full feed to a bare barcode, and turns it into complete product records built to your rules, at every level your catalog is organised by. This article says what that means in practice, who it is for, and how we start.

Who it is for

Catalogica is built for the people who run a catalog at a retailer or marketplace that buys from many suppliers: heads of ecommerce, catalog operations, content teams. Dozens to hundreds of feeds, each in a different shape, feeding one catalog and often several markets.

It is in production at Ozone.bg, a multi-supplier online retailer in Bulgaria, where more than 70,000 products have been enriched.

What happens to a product

Six steps, and every product goes through them.

Import. We connect each supplier’s feed and map it to your catalog. A new supplier is a setup task for us, not a project for you. Spreadsheets are imported in bulk, with a report of every row that did not go through.

Enrich. The AI researches the product on the open web, starting from as little as a barcode or a name. It places the product in your category tree, fills the structured specs with correct units, selects and cleans images, keeps the datasheet, and writes the title, the description and the SEO fields. When there is no usable evidence for a product, it says so instead of inventing content.

Review. Your editors open any product, change what they need, and decide what the record finally says. Whether they finish a record in our editor or in your own system after delivery, the version your editors finish is the one that counts.

Publish. As soon as a record is ready, we notify your shop, marketplace or PIM, and your system pulls the full record over the API.

Measure. Every editor correction is compared against what the AI produced, category by category. You see where it gets your catalog wrong, and so do we.

Improve. The products your team finishes become the examples the AI follows on the next run, and what the comparison shows becomes rules we write and tune with you.

Your rules, your conventions

Every retailer has conventions: how a title reads, which attributes a category needs, how a unit is written, which wording a regulated category may use. In Catalogica those conventions are rules. A rule is set once for the whole catalog and refined at any level your catalog is organised by, categories first, down to a single attribute. Whatever the supplier sent, the listing comes out built to your conventions.

Implemented by us

There is no project team needed on your side. We connect the feeds, map the data, write and tune the rules, and support your editors in their own language. Your editors review and approve. That is the whole division of labour.

What you pay for

You are billed for the AI tasks each product actually needed, not a flat fee per SKU. Researching a product, choosing its category, extracting its attributes, writing its title, its description or its SEO fields, finding its images, reading its datasheet: each is a separate, priced task, and a product that arrived with a clean datasheet does not pay for the research it did not need. There is no setup fee and no annual contract.

How we start

We typically start with a scoped pilot on one feed, so both sides see real records before expanding. Pick the supplier whose data gives your team the most work, and book a demo. We will walk through that feed together.