AI Chatbot Cannot Retrieve Product Information from Variant Codes Listed in Product Page Tables

We are experiencing an issue with our AI chatbot.

The chatbot's knowledge base contains approximately 1,500 website URLs, most of which are product catalog pages. Each product page includes several related product codes (up to 10 variants per product). These variant codes and their descriptions are displayed in an HTML table on the product page.

The problem is that when users search the chatbot using one of these product codes, the AI agent is unable to retrieve the corresponding information and does not provide a link to the related product page.

For example, on the following product page:

https://www.cappellottosrl.com/prodotti/trattamento-aria/sistemi-di-ventilazione-residenziale/air-fresh-i-s/product/D01203-AIR%2BFRESH%2BI-S/

the variant codes are listed within a table. However, when one of these codes is entered into the chatbot, the chatbot cannot identify the product, retrieve its information, or direct the user to the correct product page.

Could you please advise how this can be resolved?

Specifically:

• Does the crawler/indexing process extract data contained within HTML tables?

• Are product variant codes being indexed correctly?

• Is there a recommended way to structure or expose these codes so that the chatbot can retrieve them reliably?

• Would it be better to provide this information through a separate data source, such as structured data, a CSV feed, or dedicated product URLs for each variant?

Any guidance on improving the indexing and retrieval of product variant codes would be greatly appreciated.