Product feeds power how VWAM understands and recommends your products. The quality and structure of your feed directly impact how effectively your AI agent matches products to customer intent.
Use product feeds to:
Provide your AI agent with product data
Improve recommendation accuracy
Add rich product context (attributes, descriptions, pricing, etc.)
Control which products your agent can recommend
Before you begin
Before uploading your product feed:
Gather your product data in CSV format.
Identify required fields (ID and Title).
Determine which additional attributes (size, color, materials, technical specs, etc.) help your AI agent better understand your products.
Decide whether you are creating your first feed or uploading additional feeds.
Note: The template is optional but recommended for first-time uploads.
Create a Product Feed
Download a template
The template provides a suggested structure, but you can add new columns. The additional columns are what give the additional context to the AI. For example, if there are certain products that have certain color or sizes, you may want to include these columns in your feed if they are useful information for the AI to decide on.
Populate your feed
The more data you provide, the better VWAM performs. There are technically only 2 required fields in the feed which are the ID and the Title. However, there are 6 consumer facing fields which are eligible to show to end users.
Strong feeds typically include:
ID
Title
Description
Price
Image URL
Brand
Category
Redirect URL
Key Attributes; (size, color, material, features, etc.). These are unlimited and have no structure
Upload your feed
Click + New Feed.
Select your completed CSV file.
Upload the file.
VWAM confirms when the upload is successful.
Link your feed to an agent
After uploading your feed:
Click into your agent from the Dashboard.
In the Editor, locate the Collections section.
Assign your feed to your agent.
Best practices to maximize AI performance
1. Write rich descriptions
Include benefits, use cases, and differentiators — not just short blurbs.
Weak: “Men’s Running Shoe”
Strong: “Lightweight men’s running shoe designed for long-distance comfort with breathable mesh upper and responsive cushioning.”
The AI uses this language to understand intent matches.
2. Include Structured Attributes
Add attributes as additional columns in your CSV when applicable:
Size
Color
Style
Fit
Technical specs
Materials
The more structured and descriptive your data, the more precise your re
Next steps
Test your agent to confirm products display correctly.
Create an agent and assign your product feed in the Collections section to begin recommending products.
Review product recommendations for accuracy.
FAQ
Is the feed automated?
Not at this time. We are adding automation later this year.
Are all fields required?
No. The only required fields are ID and Title. All other fields are optional. However, more information leads to better recommendation quality.
Can I upload multiple feeds?
Yes. You can upload as many feeds as needed and switch between them. You can also rename your feeds by clicking on the three dot menu.
