Semantic search is a technique that enables search queries in terms of the user’s intended meaning and context rather than simply matching keywords.

Search terms are converted to vectors in a high-dimensional space and compared to representations of items in the shop, enabling the presentation of items that match the meaning of the search term, but have a different name and ultimately helping the customer find items they might otherwise have missed.

The vector database forms the backbone of the semantic search engine. It is capable of storing vectorised representations of e-commerce items that are subsequently compared to search terms passed by the Assistant.

The comparative distance in multi-dimensional space determines the ranking in which items are returned back to the Assistant and presented to the Customer, ensuring that the items with the highest relevancy towards the conversational context are displayed first.

A Semantic Search embedded within the Conversational Assistant drives sales by enabling a product presentation according to the meaning of the customer and not the word representing the search term.

For example, a semantic search returns the item 'Rocket Salad', even when the customer searches for a different name of the same product, such as 'Arugula'.

In this regard the Conversational Assistant can be compared to a physical Sales Associate with expert knowledge on what the Customer is trying to achieve with the products they are looking for and what you have in stock.

We offer a vectorisation pipeline to set up a Semantic Search Engine with minimal effort .

As a fully managed SaaS service, we subsequently take care of hosting, maintaining and continually updating the vector database for you, so that Assistant continues to guide your customers in a powerful product exploration experience.