Think about how much has changed over the last two years: Around this time in 2024, Charli xcx was gearing up for a “Brat summer.” Paris was preparing to host the Olympics. And consumers still shopped by doing Google searches and visiting websites.
Fast forward to today, when people are almost as likely to ask an LLM or AI agent to do their shopping for them. Salesforce research shows that 39% of consumers have already used AI to find products. And last holiday season, Salesforce estimated that AI- and agent-referred traffic drove 21% of all holiday orders globally, adding up to $263 billion in sales.
“We’ve seen rapid adoption of consumers using third-party AI tools like ChatGPT, Perplexity, Gemini, and Claude to help them find products,” said Caila Schwartz, director of industry insights at Salesforce. “These tools are revolutionizing the consumer shopping journey.”
This presents a new challenge for businesses. If artificial intelligence (AI) agents are helping with the shopping, how do you make sure they’re finding — and recommending — your products? Let’s take a deep dive into agentic commerce, and offer a few tips to make sure your brand stays front and center.
How big a deal is agentic commerce? A 2025 McKinsey report called it “a seismic shift in the marketplace.” The consulting firm also compared it to the e-commerce boom, but one that would happen much more quickly.
Little surprise, there. Agentic commerce has one huge advantage over e-commerce: It drills down on the shopper’s intent. In other words, what is the consumer really looking for? In the past, someone might have searched Google for a “warm jacket.” Now, they can tell AI they’re going to Vancouver for the weekend, it’s going to be 40 degrees Fahrenheit, and they need a well-insulated jacket that’s stylish enough to wear in a city. The results are no longer based on what keyword rankings elevate. They identify items that most closely match what the shopper is looking for.
Salesforce research has found that people use LLMs and AI agents to save time, find the best deals, and receive personalized product recommendations. And the good news is that shoppers who arrive at retail websites from generative AI platforms are two times more likely to buy something than those who arrive from other sources of traffic.
But even more, agentic commerce is poised to open doors for companies. “ChatGPT and other LLMs are broadening the pool of customers, or potential customers, that could actually reach your brand,” said Pei Thong, product marketing director at Salesforce.
With everything moving so quickly, businesses need to tailor websites and content not just for human readers, but also for AI. They need to design systems that are visible, trustworthy, and attractive to both audiences. “The challenge for brands now is, how do I connect my brand into that space?” said Schwartz.
Here are three practical tips to get you started:
Research shows that 50% of business owners and marketers worry their products won’t appear in AI search results because of poor data. They’re right to be concerned. AI needs well-structured, accessible data to find and recommend your products.
“You need to make sure your catalog is super clean and accurate, and your website is up to date,” said Thong. Your catalog should include structured data that tells LLMs exactly what an item is: the brand, model name, SKU, price, color, size, and dimensions, for example. Be sure that the “About Us” section on your website and your shipping and return policies are easy to find and clear.
But also include less-structured data that matches how shoppers search. Use words such as “waterproof,” “vegan leather,” or “BPA-free.” And offer compatibility suggestions, such as “this jacket is perfect for chilly days” or “this shirt pairs well with our wide-legged jeans.”
To give you an idea of how specific your product descriptions should be, consider this example from the beauty world: “This 10% Vitamin C Serum is formulated with pure L-ascorbic acid and ferulic acid at a pH of 3.0 for maximum stability. It targets hyperpigmentation and fine lines. Suitable for all skin types, it is fragrance-free, paraben-free, and non-comedogenic.”
This level of detail doesn’t just inform the shopper. It also gives AI the data it needs to make better recommendations.
Once you’ve cleaned up your data, your next step is to integrate your catalog into LLMs.
Salesforce B2C Commerce customers can do this through two new direct catalog feeds for OpenAI and another for Google. The OpenAI Agentic Commerce Protocol (ACP) integration is currently in beta and will be widely available the summer of 2026. And the Google version — Universal Commerce Protocol (UCP) — will be available this summer, too. Both are open source standards that allow information to flow between buyers, AI agents, and businesses.
Here’s how it works: After companies turn on the feature, their catalog data will automatically populate in the integration fields. Businesses can correct fields as needed, and set automated updates at a frequency that works best for them. And once the integrations are live, they will make products easier for ChatGPT or Gemini to find.
But there’s another advantage to integrating your catalog. “We know that LLMs are prioritizing structured data — specifically product feeds — over web scraped results,” said Thong. That means more eyeballs on your products and more shoppers ready to buy.
“AI needs to trust your brand if they’re going to recommend it,” said Schwartz, “so it’s not just about how you format your website and the data that you surface, but also what’s happening off your website.”
Put simply, you need third parties to review your products. When agents search for products, they not only visit company websites to validate information. They also look at reviews to see what people think of the product. Think TikTok reviews, earned media mentions, or comments on Reddit.
If a product receives a large number of positive reviews, LLMs consider that as a sign of its quality. And if 100 people on Reddit recommend your site for the “best eco-friendly sneakers,” LLMs are significantly more likely to recommend you in a chat response.
Schwartz predicts agentic commerce will change how marketing departments operate, and that public relations may step back into the spotlight, particularly in their outreach to media and influencers. “Influencers are going to probably be even more important, as is the micro-influencer trend. It’s not necessarily about how many followers an influencer has, but are they engaging the right audience for your product? Is their content authentic and is their community highly engaged? That’s going to be really important,” said Schwartz.
Your most powerful asset, though, may be your community of customers. Ask customers to review products they’ve purchased. Send them personalized emails that encourage responses. Engage them with your brand.
When customers actively participate with a brand, it sends a strong signal to LLMs that people care about the products. At the end of the day, said Schwartz, “it’s all about building trust.”
If there’s one weird blip in the rise of agentic commerce, it’s that Gen Z shoppers prefer brick-and-mortar stores. PwC found 61% of Gen Z now prefer to discover new products in-store. They’re looking for the experience and immediacy that only in-store shopping offers.
But the shopping journey tends to be complex and disjointed. Salesforce research found that the average shopper interacts with a brand nine different times before making a single purchase. In 2025, 80% of consumers made at least one purchase in a store, 83% made a purchase online, and 24% through a social media platform.
Social media also plays a big role in how Gen Z decides on products. YouGov found 35% of Gen Z shoppers discover products through influencers or bloggers, and 64% use social media to research products.
The takeaway: It pays to take an omnichannel approach. Create an experience in your stores. Build relationships with social media influencers, and encourage community with your customers. Make sure LLMs can find — and recommend — your products.
It’s still early days for agentic commerce, and most companies are only now starting to think about how they can make their products easy for AI to find.
But if there’s one thing everyone can agree on, it’s that you need to meet your customers where they are. “And a lot of times,” said Thong, “that’s on shopping channels like Gemini or ChatGPT.”
It is a new way of shopping where Large Language Models (LLMs) or AI agents find, research, and recommend products for consumers.
About 39% of consumers already use AI to discover new products.
Shoppers coming from AI platforms are two times as likely to buy something than other visitors.
You should clean your data, integrate your catalog into AI tools, encourage third-party reviews, and build community with your customers.
Yes, AI agents visit websites to validate information and check details like shipping policies.
Yes, AI agents look at reviews on sites such as TikTok and Reddit to decide if a brand is trustworthy.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。