Nosto launches AI Search, with hybrid keyword and vector matching live now

Why Nosto is focusing on search

The company argues that search is where shopper intent turns into sales or gets lost. Its example: a shopper typing "repair conditioner" and another asking for "something for hair that feels dry after washing" may want the same product.

Nosto cites its own Censuswide research with 2,000 US and UK consumers to back this up. According to the company, 50% said a poor search experience had stopped them buying, even when they arrived intending to purchase. 63% said they were more likely to return to a store whose search helped them find what they wanted easily.

The research also points to a language gap. Nosto says 62% of shoppers often know what they are looking for but struggle to find the words to describe it. It adds that 72% of respondents have already used AI search features on a shopping site or are open to trying them. Experiences such as ChatGPT, Claude and AI Mode in Google Search, the company says, are raising expectations for how easily people can find products on a retailer's own site. When onsite search returns irrelevant results, 40% say they are likely to leave and ask a tool such as ChatGPT or Claude for recommendations.

What is available now and what is coming

The product that ships today is Hybrid Vector Search. It combines two approaches: keyword search, which matches the terms a shopper enters, and vector search, which interprets semantic relationships between a query and the product catalog. Nosto says this helps surface relevant products even when the wording differs from the catalog, while still supporting searches for specific brands, product names or references.

Three capabilities are marked as coming soon, all aimed at reducing the effort of narrowing results. Nosto says 61% of shoppers found that the process of finding what they wanted often took too long.

• Attribute Recognition: search will identify supported attributes such as color or price and apply the matching filters. For "conditioner for dry hair under $40," the budget would be applied without the shopper re-entering it in the filter panel.
• Query Refinements: search will offer follow-up choices, such as "Is your hair fine or thick?", so shoppers can pick an option instead of rewriting the query.
• AI Summaries: short messages will explain how search interpreted a request and how the selection relates to it.

Personalization, merchandising and conversational mode

AI Search builds on Nosto Personalized Search and keeps its personalized ranking and merchandising controls. Nosto gives an example: a haircare retailer can use merchandising rules to boost relevant conditioners with high stock levels, while personalized ranking gives more visibility to products from the shopper's preferred brands.

For shoppers who want more guidance, Nosto is also offering AI Mode in early access. It acts as an entry point into a conversation with Discovery Agent, which the company describes as its separate conversational discovery product. Shoppers can ask follow-up questions and explore recommendations, such as complementary products for a haircare routine, while those ready to buy can continue through the search results.

What it means for retailers

For ecommerce teams, the pitch is to extend product discovery within an experience they can still shape, rather than replace their existing merchandising setup. Nosto says effective search can remove barriers to purchase and give shoppers fewer reasons to continue their search elsewhere.

Interested retailers can book a demo, and existing Nosto customers can select AI Search in a request form to be contacted by their Nosto team. The company is also hosting a webinar on November 12 covering the research and the product. Note that the consumer figures come from Nosto's own commissioned research, and only Hybrid Vector Search is live today.