Product search has gone from panoramic to personal. Prospective customers want their recommendations to incorporate previous purchase history, account for current behavior, and predict potential needs.
Agentic commerce powers this personalization, making it possible for brands to tailor product suggestions for every search and every shopper.
And with 51% of holiday shoppers saying they're comfortable with using agentic AI in Salsify's "2026 Holiday Pulse Report," it's not something that can be overlooked.
Here’s an examination of the ongoing shift in AI product search, explaining how AI agents make this all possible, evaluating the benefits of better suggestions, and exploring strategies for personalized success.
Keywords powered the first generation of product search, and, as a result, quantity rather than quality became the driving force of search engine optimization (SEO). To address this issue, search engines adjusted their algorithms to consider content and context along with keywords.
This paved the way for search segmentation. Brands began to leverage demographic and purchasing data to define customer segments — groups of customers that shared similar purchasing priorities, budgets, or locations — and used these segments to create tailored marketing and SEO campaigns.
While this enhanced the overall search experience, challenges remained. Thirty-three percent of customers still report that their top search frustration is being offered products irrelevant to their needs, according to Salesforce.
The rise of AI set the stage for a new approach: personalization. Instead of viewing the customer as one of many or one of a smaller subset, personalization prioritizes consumer-specific search habits, purchasing histories, and potential needs.
According to Salesforce, 65% of customers say they will stay loyal to companies that offer a more personalized search experience.
And Salsify's "2026 Consumer Research" report found that 25% of shoppers will trust an AI recommendation enough to buy if there is a personalized explanation for why it fits their needs.
AI combines data from multiple sources to create a holistic view of customer preferences.
Consider a customer searching for a roof repair company. Using traditional search, they input the keywords “roof repair” and hope that sites with great SEO are also reputable and cost-effective.
A segmentation-based approach sees companies crafting more tailored content that they attempt to match with interested customer segments. This means a roofing company might create multiple landing pages, with each one targeting a different town or neighborhood to help ensure content reaches their target audience.
AI goes a step further, using a combination of company and customer data to create personalized search results. In the case of our roofing repair service, AI could combine keyword and location data with other factors such as previous purchases and the likely age of a customer’s home based on their neighborhood.
Rob Gonzalez, Salsify co-founder and chief strategy & innovation officer, has personal experience with using AI to power personalized searches for his family. As he wrote in his whitepaper, "I use Claude for my personal life, and it has access to a personal document I’ve loaded up ... So, when I asked, "We’re done with the cheap umbrellas – find us one that is the right size but can actually withstand her," I didn't have to explain the why, the “context.” The agent already had it."
Using this information, AI-enabled search can deliver buyer-specific offers that account for current needs and prospective concerns.
Agentic commerce fuels the next iteration of this effort: hyper-personalization. Research firm Deloitte defines hyper-personalization as the use of data, analytics, AI, and automation, which allows companies to send “highly contextualized communications to specific customers at the right place and time, and through the right channel.”
Where AI agents analyze data sources to discover new patterns, they can extrapolate context from content, allowing AI tools to better understand what users are really asking and provide net-new answers beyond traditional search.
In the example of our prospective roofing customer, this could mean leveraging contextual data, such as recent weather and news reports, that inform consumer intent.
For example, if the county has just experienced a string of powerful storms, customer repair needs may be substantial. Instead of sending users to a generic home page or contact form, agentic AI points customers to services that meet their specific needs.
Personalized (and hyper-personalized) product searches offer multiple benefits for brands.
Retail giants like Amazon use agentic AI to create personalized product recommendations. Amazon COSMO (Common Sense Knowledge Generating and Serving System) analyzes shoppers' purchase histories, browsing histories, query histories, and brand engagement to determine search intent.
For example, if a shopper searches for a "denim jacket" on Amazon COSMO, the AI can determine what kind of jacket they're looking for based on their behavior data.
Agentic AI can also go beyond product pages and text descriptions to provide multimedia experiences. This could include relevant explainer or comparison videos for complex products, or the use of audio files for products that depend on sound, such as musical instruments or speakers.
Highly personalized marketing is highly effective, but comes with a high cost.
While segmentation takes more time and effort than a simple keyword-based search, it still allows companies to group customers into similar marketing buckets.
For example, a company with 10,000 customers might have 10 buckets of 1,000 customers. Each bucket gets its own marketing effort, meaning the team needs to create 10 targeted campaigns that deliver tailored search results.
Using a personalized approach, companies need 10,000 individual marketing efforts — a task that even the best teams may struggle to complete on time and budget.
Using AI streamlines this process. Brands can simply provide AI with the details and let algorithms create personalized search results based on user input.
Personalized recommendations drive more conversions. According to Salsify and the Digital Shelf Institute’s “Ecommerce Pulse Report: Q4 2024,” 37% of shoppers say they buy more because of personalized product suggestions.
Given that 52% of shoppers review two to three channels before making everyday purchases like food and beverage, according to Salsify’s “2026 Consumer Research” report, companies are best served by capturing consumer attention ASAP.
While AI underpins personalized results, four strategies can help boost intelligent tools’ impact.
Agentic AI improves over time through data — the more data available for analysis, the better the results. To set AI up for success, brands need to provide large amounts of data, ensure it’s accurate, and measure the output of AI against applicable ecommerce key performance indicators (KPIs).
As AI’s capabilities expand, more resources are required. Brands need a plan to scale personalized search without compromising user experience or speed.
IT infrastructure lays the groundwork for search success. While there’s no single way to invest — opting for the capital expense of on-site servers or the ongoing operational costs of cloud resources are both viable options — brands must budget for sustained IT investment.
AI is powerful because it can combine disparate data sources into new insights. But this also presents potential privacy and ethical risks. Two rules can help brands avoid ethical issues.
First, be upfront and clear about AI use. Tell consumers exactly what data is collected, how it’s used, and what they can expect. Give them a clear option to opt out.
Second, use secure AI tools. This could mean building in-house solutions that live within corporate networks or leveraging enterprise-grade tools that offer enhanced protection. Under no circumstances should brands input customers’ data into public-facing AI tools.
AI-powered product search is a powerful tool. But AI is a work in progress, meaning that despite best efforts, it still gets things wrong — confidently wrong. To ensure outputs match expectations, keep a human in the loop. Have staff regularly check AI results for accuracy, relevancy, and completeness.
Brands have precious little time to capture and keep customer attention. Tailored AI product searches offer a way to thread the needle — to walk the line between organic outputs and personalized results that keep buyers returning.