Prepare useful product knowledge
Give the assistant clear catalog information, policies and FAQs—and verify what it actually learned.
Source quality is a practical part of being AI-ready. Start with information your store already stands behind: product descriptions, variant details, published policies and answers to common questions.
Choose and review sources
Use the source types available in your installed version. Do not assume that every product, page or file was ingested just because a training action completed. Review the reported coverage and test representative products, especially in larger catalogs.
Avoid contradictory policies and outdated offers. Keep private staff notes, customer information and credentials out of public knowledge sources. Only provide content you have permission to use.
Train a small set first
Check your Gemini connection, add a bounded set of sources, and run the available training process. Review failures before expanding the set. If source updates need reprocessing, complete that step before expecting new answers.
Verify instead of assuming
Ask about a specific product characteristic, a variant, and a returns scenario. Compare each answer with the source. Also ask something the sources do not answer and check that the assistant does not invent a policy.
Product knowledge in Getail is not automatically a public feed to third-party AI agents. Merchant Center distribution is a separate configured workflow; see local availability.