How Will AI-Assisted Shopping Impact Retail Media and Commerce?
This week, advanced artificial intelligence platform Perplexity launched its AI-powered shopping assistant, which allows users to research and purchase products all within its website or app. What does this mean for the future of commerce, and how will it impact shopper behavior? What will be the impact on the retail media ecosystem or influencer marketing? How do you advertise to an AI agent? I reached out to Jason Goldberg, chief commerce strategy officer at Publicis Groupe, to get his take on the “Shop Like a Pro” launch. Here’s how he predicts AI-assisted shopping will transform the world of commerce …
P2PI: How do you think the introduction of AI shopping assistants, like Perplexity’s Shop Like a Pro, will change consumer behavior compared to traditional e-commerce methods?
Goldberg: The jury is still out. Historically, many shopping journeys have started with Google Search, but tools like Perplexity — and even Google’s own AI-powered Search Summaries — represent a significant evolution. These AI-driven solutions are poised to become better mousetraps for certain use cases. Features like Shop Like a Pro will undoubtedly integrate into the broader shopping ecosystem.
The bigger question is whether these AI assistants will fundamentally reduce the reliance on traditional search altogether. For example, will grocery replenishment shift entirely to [Amazon’s] Alexa? Will [Apple’s] Siri choose the perfect gift for you instead of requiring hours of research? Or will other AI agents bypass traditional interfaces altogether and operate via APIs instead of HTML? These possibilities are part of the future, but the scale and speed of adoption remain unclear.
P2PI: What are the implications for advertisers as AI agents begin to handle shopping decisions? How might this shift how brands and retailers approach targeting and advertising strategies?
Goldberg: The rise of AI agents could profoundly disrupt advertising. Today, much of advertising focuses on capturing consumer attention and targeting buying intent through lower-funnel tactics, such as search ads and retail media. But what happens when a consumer delegates pantry replenishment to an AI agent instead of searching for peanut butter on Instacart? Where does the peanut butter advertiser place its ad in that scenario?
This could lead to a significant shift toward top-of-funnel brand building. Even in search, the transition to large language model (LLM)-driven algorithms complicates traditional keyword strategies. Currently, we understand what factors drive rankings on platforms like Google, Amazon or Walmart. With LLMs, even their creators may struggle to explain why certain results surface. Additionally, these algorithms may tailor criteria for each user, making the process highly dynamic and less predictable. Marketing strategies in an AI-driven world will require entirely new playbooks.
P2PI: How might this AI shopping experience impact the broader retail media ecosystem? Will traditional display ads and search ads lose relevance, or will new formats of retail media emerge in this space?
Goldberg: AI agents will likely reduce explicit product searches and shift purchasing toward implicit, needs-based fulfillment. This could decrease the number of impressions available for retail media network (RMN) advertisers. With more persistent product and brand preferences, the first purchase could become a high-stakes, winner-take-all scenario for advertisers.
AI agents could open doors for new advertising opportunities in the form of hyper-personalized advertising within their interfaces, where ads are tailored to the agent’s understanding of user preferences, intents and historical data. The unknown question is if new AI interfaces emerge or if AI simply causes the interface to fade into the background.
The bottom line: Advertisers may face higher costs as they compete for fewer, but more valuable impression opportunities. Retail media will need to adapt, potentially emphasizing new ad formats or platforms better aligned with AI-driven shopping experiences.
P2PI: Do you foresee AI agents replacing some of the current “influencers” or product review platforms? If so, how might this affect content creators and affiliate marketing?
Goldberg: It’s possible, but unlikely. AI agents are more likely to leverage influencer-generated content to inform their recommendations rather than replace it. For instance, an AI engine might only know that a particular dress has a fantastic “hand feel” because an influencer described it that way. Content creators will remain an essential part of the ecosystem, but their role may evolve to focus on feeding high-quality insights into AI recommendation engines. AI agents likely will diminish the individual brand value of influencers by aggregating content into easier to consume summaries.
P2PI: With the potential for AI agents to take over much of the shopping research and purchase decision-making process, how do you see the role of traditional retail websites and advertising changing over the next 5-10 years?
Goldberg: It’s likely we’ll see fewer consumers visiting retailer websites to shop. Instead, digital retail will evolve to prioritize publishing commerce capabilities — like product data, order management and transaction services — via microservices. These capabilities will then be consumed by various AI agents, many of which we haven’t even imagined yet.
Retailers may no longer “own the glass” through which consumers shop, shifting their focus to enabling seamless integrations with these agents. Advertising will also need to adapt, emphasizing visibility in these AI-driven ecosystems rather than relying solely on traditional website traffic. Of course, retailers’ websites are still going to be important for a subset of shopping missions that are more focused on discovery, storytelling and fostering trust. We used to say that e-commerce solved buying, but broke shopping. A pattern may emerge here where the AI agents take over routine buying and replenishment at first, while retailers still own discovery and inspiration.
P2PI: What do you think will be the main challenges for large e-commerce platforms such as Amazon and Walmart as they integrate AI into their shopping experiences?
Goldberg: Some of their existing strengths, like logistics and vast product assortments, will continue to serve as competitive moats. However, new challenges will emerge. The old adage, “What got us here won’t get us there,” has never been more relevant.
Retailers such as Amazon and Walmart are essentially aggregators, making it convenient for shoppers to purchase goods from multiple providers. AI agents, however, could disrupt this aggregation model by connecting consumers directly to suppliers. In a future where AI eliminates some of the advantages of scale and convenience, retailers may struggle to profit from selling third-party products. Of course they also have some tremendous advantages like their customer base, data, infrastructure and ecosystem, but I wouldn't bet against them being able to leverage those advantages moving forward.
P2PI: What’s your prediction on how quickly consumers might adopt AI assistants into their regular shopping journeys?
Goldberg: Adoption will likely follow the classic trajectory: “slowly, then all at once.” In the short term, current tools like Perplexity’s Shop Like a Pro are still too slow and clunky for everyday use. The volume of commerce-related traffic running through these AI systems today isn’t yet significant, and user experiences need refinement.
However, high-volume use cases will arrive faster than most people expect. We’re likely in an abbreviated hype cycle for AI agent commerce. While we’re still climbing the “peak of inflated expectations,” the industry could reach the “plateau of productivity” within the next 24 months.