Skip to main content

Sales Catalog

Overview​

The Sales Catalog is where you keep the products and services your team actually sells: name, description, brand, category, variants, prices, initial stock, and images. This is distinct from the Catalogs tab inside Aurora AI (which is meant for training the agent with reference data): this catalog is your business's commercial inventory. Once configured, you can connect it to one or more AI agents so they answer customer questions with real product information — availability, price, attributes — without having to upload products as documents.

How to Access​

  • Navigate to Tools > Sales Catalog (also available directly at /sales).
  • To browse and read: any team member role.
  • To create/edit products, import, and configure: Administrator.
  • To connect the catalog to an AI agent: Chatbot Admin.

Interface Overview​

The catalog is organized into two main areas:

  • Products list (/sales/products) — The default screen: a table with every product and service, category filters, SKU search, an Add product button, and the New import button for bulk uploads.
  • Catalog settings (/sales/settings) — The configuration hub. Has tabs for Categories, Custom properties, Import templates, and Import history.

Each product has its own detail page (/sales/products/:id) where you manage variants, attributes, images, and pricing.

Key concepts​

Product vs. Service​

A product represents something physical you sell with stock; a service represents something intangible or bookable (consulting, a session, an appointment). The type controls which flags and fields show up when you edit it.

Variants​

Every product has at least one default variant with its own SKU, price, and currency. If you sell the same product in different presentations (sizes, colors, flavors), you add additional variants — each with its own SKU and price.

Categories​

Products are organized into a category hierarchy (with parents and children). Categories help you navigate the catalog, and they also let you scope what an AI agent can see: you can restrict an agent to a branch of the tree and all its descendants.

Custom properties​

Beyond the standard fields (price, brand, description), you can define your own typed properties — numbers, text, checkboxes, dates, multiple choice — to capture attributes specific to your business (weight, color, material, voltage, launch date). These properties are what lets the AI agent filter precisely on "under 10 kg" or "color blue".

When an AI agent queries the catalog, it doesn't use simple keyword matching: it combines semantic understanding (it knows that "jacket" and "coat" can mean the same thing) with structured filters (price, brand, category, properties). That means customers can phrase things naturally and still get precise results.

Important Notes​

  • The catalog is per-account: each Aurora account has its own isolated catalog.
  • There is no hard cap on the number of products in the catalog (unlike the 3,000-item limit of the Aurora AI Catalogs). There is a maximum of 10,000 rows per import file.
  • Per-agent activation is opt-in: configuring the catalog does not expose anything to customers until you connect the catalog to a specific agent.

FAQ​

Q: How is this different from the Aurora AI Catalogs (Knowledge Base)? A: Aurora AI Catalogs are generic data sources the agent consults alongside your documents. The Sales Catalog is your business's structured inventory, with variants, prices, and stock, and connects to agents through a dedicated integration with hybrid search.

Q: Do I need to activate the catalog for every agent? A: No. Each agent connects individually. You can have several agents and choose which ones consult the catalog — even restrict each to specific categories.

Q: Do my customers see catalog prices and data directly? A: Only when a connected AI agent answers their questions. The catalog itself is not publicly exposed.

Q: Can I import from Excel, CSV, or a PDF/Word document? A: Yes. The importer supports three formats: CSV templates, free-form column mapping (Excel/CSV), and AI-assisted extraction from PDF, Word, or images.