Docs/Smarter AI/How Product Recommendations Work

How Product Recommendations Work

Understand how Kwiro reads your catalog and matches the right products to each customer's question.

How Kwiro Finds the Right Products for Every Customer#

When a customer asks "I need a warm jacket for hiking," Kwiro does not just search for the word "jacket" in your product names. It understands the meaning behind the question and finds the best matches from your entire catalog. Here is how that works, in simple terms.

Shopper questionPlain English, 58 langsUnderstand intentSale? Gift? Compare?Search catalogVector + keywordApply memoryWhat worked beforeCompose answerBullets + cards~50ms~100ms~200ms~50msStreaming
Diagram: The recommendation pipeline. Each box is metered so you know where the milliseconds go. The whole loop runs in well under a second for a typical shopper question.

Step 1: Kwiro Reads Your Entire Catalog#

When you first install Kwiro, it reads every published product in your WooCommerce store -- names, descriptions, prices, categories, tags, images, stock status, ratings, and reviews. It builds a deep understanding of what you sell.

Whenever you add, update, or remove a product in WooCommerce, Kwiro automatically picks up the change. You never need to manually re-sync.

Step 2: A Customer Asks a Question#

A shopper opens the chat widget and types something like:

  • "What's a good gift for a 10-year-old?"
  • "Show me red dresses under $80"
  • "I need something waterproof for camping"

Step 3: Kwiro Understands the Request#

The AI breaks down the question to understand what the customer really wants. For "warm jacket for hiking," it identifies:

  • Product type: jacket or outerwear
  • Feature needed: warm / insulated
  • Use case: hiking / outdoor activity
  • Implicit needs: probably durable, possibly waterproof

Step 4: Finding the Best Matches#

Kwiro searches your catalog for products that match the customer's needs -- not just by keywords but by meaning. A product described as "insulated trail coat perfect for cold-weather adventures" would match even though it does not contain the word "jacket" or "hiking."

It considers price range (if mentioned), stock availability (never recommends out-of-stock items), customer ratings, and how well the product fits the request.

Step 5: The Customer Sees Recommendations#

Kwiro responds with a helpful message and shows product cards -- visual tiles with the product image, name, price, star rating, and a "View Product" button that links directly to your product page.

The AI typically shows 1-3 product recommendations, ranked by how well they match. It explains why each product is a good fit, so the customer feels confident about their choice.

Why Good Product Descriptions Matter#

Kwiro is only as good as the information in your catalog. Products with detailed descriptions, accurate categories, and relevant tags get recommended more accurately. If a product has a one-word description, the AI has very little to work with.

Take a few minutes to review your top products and make sure their descriptions mention key features, materials, use cases, and who the product is best for.

Was this page helpful?
Updated April 2026

Related reading