AI Product Recommendations

Automatically link products based on their attributes, descriptions, and images. The AI analyzes your catalog to find the best matches for cross-selling, alternatives, or complementary accessories.

[try:/products|Open Products to Start|shopping-cart]

How it Works

The AI uses a two-step process to create relationships:

  1. Candidate Selection: It searches your catalog for potential matches based on your instructions.
  2. Semantic Matching: It compares the source product with the candidates to find the most relevant links.

You can use this feature to build complex product relationships without manual searching.

Relationship Types

Choose the relationship type that fits your business goal:

card: Related Products
icon: link
Find products that complement the source item. Use this for cross-selling (e.g., shoes and socks) or similar items in the same category.

card: Alternative Products
icon: repeat
Identify direct substitutes or upgrades. Perfect for suggesting replacements when an item is out of stock or offering a premium version.

card: Product Accessories
icon: plus-circle
Link add-ons that enhance the main product. This includes cases, chargers, or maintenance kits that work with the specific item.

card: Set Items & Bundles
icon: package
Create product kits or bundles. The AI can select multiple items that belong together and even suggest quantities for each.

Generating Recommendations

You can generate recommendations for a single product or in bulk.

Using Mass Update (Bulk)

Process hundreds of products at once using the batch wizard.

1. Select the products you want to process in the grid.
2. Click **Mass Update** [badge:zap|info] and switch to the **AI-assisted smart update** tab.
3. Select **Generate Relations** as the operation type.
4. Choose the destination relation field (e.g., Related Products, Alternatives).
5. Enter your instructions. Be specific about what kind of relationship you want.
6. Click **Generate Preview** to see the suggested links.
7. Review the matches and click **Apply** to save the relationships.

Using Field AI (Single)

Update relationships for an individual product directly from the form.

1. Open a product and find the relation field you want to populate.
2. Click the **AI** button next to the field.
3. Describe the relationship you're looking for.
4. Review the suggested products in the preview.
5. Click **Apply** to link the selected items.

Example Use Cases

Use these prompts as a starting point for your own recommendations.

1. Cross-Sell Complementary Items

Link items that customers often buy together.

{
  "prompt": "Find products that complement this running shoe, such as athletic socks, insoles, or water-repellent sprays.",
  "fields": ["relatedProducts"]
}

2. Suggest Direct Alternatives

Find substitutes with similar technical specifications.

{
  "prompt": "Suggest alternative shipping boxes with the exact same dimensions (width, height, depth) as this one.",
  "fields": ["alternativeProducts"]
}

3. Build Product Bundles

Create a starter kit from individual components.

{
  "prompt": "Create an office starter kit bundle including a desk lamp, wireless mouse, and a large mouse pad.",
  "fields": ["setItems"]
}

Tips for Better Results

card: Use High-Quality Data
icon: database
The AI relies on product names, descriptions, and attributes. Ensure your source data is accurate for better matching.

card: Be Specific
icon: target
Instead of "find related items", try "find accessories that are compatible with this specific model and brand."

card: Leverage Images
icon: image
If your products have images, the AI can use them to identify visual similarities and complementary styles.

card: Review Candidates
icon: eye
The AI only picks from the candidates it finds. If the results are poor, try broadening your search criteria in the prompt.

Troubleshooting

No recommendations found

  • Check if the destination entity has enough products to match against.
  • Ensure your prompt isn’t too restrictive.
  • Verify that the source product has enough descriptive data (name, category, etc.).

Irrelevant matches

  • Refine your prompt to exclude certain categories or brands.
  • Be more specific about the technical requirements (e.g., “must have the same voltage”).
  • Use the preview step to edit and guide the AI’s logic.

Wrong quantities in bundles

  • Explicitly state the required amounts in your prompt (e.g., “include 2 of each cable”).
  • Edit the quantities directly in the preview table before applying.

[try:/products|Start Generating Recommendations|bot]