“Conversational commerce” is the current flavor of agentic commerce that’s taking center stage. Using genAI systems, consumers are researching, discovering, and purchasing products and services. Recent moves from major players in the market are embedding both commerce and conversation into more experiences, expanding support for specific use cases, and offering personal AI shoppers that seem to be blowing up overnight. Let’s take a look at some notable recent moves.

YouTube Turns Content Into A Commerce Interface

YouTube is making it easier for eligible creators to tag products in content for a global audience and to earn commissions on related sales. At the same time, the watch page is becoming an interactive environment. YouTube announced a new conversational search feature that will answer questions about products in detail. In this expansion of the “Ask YouTube” conversational experience, viewers will be able to ask questions directly on a video’s watch page.

Google’s Protocol Is Growing Up In Public

Google’s Universal Commerce Protocol (UCP) continues to evolve, incrementally tackling all the messy realities of retail. It already has added support for retail and grocery-specific requirements (like operating hours for pickup locations and fractional quantities, which accommodate items sold by weight, for instance). Google promises specifications for food and lodging soon. Major updates are not backward-compatible, meaning merchants must keep their data feeds updated (or work with partners who do so).

Personal Shopping Agents Force The Access Question

Personal shopping agents are also testing who gets to mediate a retailer’s customer relationship. Some retailers’ recent updates to their terms cover them against responsibility for unintended agent-initiated purchases — by placing the responsibility with the customer. There are tugs-of-war between answer engines like Perplexity, marketplaces like Amazon, AI agents like Meta’s Muse, and individual brands over the access and actions that AI agents can have/make on behalf of customers. Most recently, Amazon moved to bar access to Muse from shopping Amazon on behalf of customers, although Amazon recently lost its similar injunction against Perplexity.

Meanwhile, Meta’s Muse tops Apple’s App Store chart, while the founder of Instinct (an AI agent launched just last year) raised $1 billion in its latest funding round. Finally, Shopify just announced that it’s enabling browser-based AI agents to purchase directly on Shopify merchants’ sites, within a day or so of OpenAI introducing dots, its own personal AI agents.

What Digital Leaders Should Do Now As Commerce Becomes More Conversational

Conversational commerce is going to be a slow transition from search results to responses, from queries to conversations, and from clicks to … well, no clicks, potentially.

Some customers may adopt conversational shopping faster than your governance teams expect or slower than your leading-edge trendsetters hope. Digital leaders should prepare for a journey in which conversations will replace static results and traditional flows by:

  • Assigning an owner for agentic access. Bring commerce, security, legal, marketing, and operations together to define which agents may discover, compare, reserve, purchase, and service orders. Determine whether to allow scraping bots to access and retain product content and whether to block or allow bots from specific sources. Allow for different answers at different times, such as allowing scraping bots just before a major shopping event so they capture the new data but not during the event, when they can interfere with human buyers.
  • Preparing product content for questions, not just clicks. Think beyond attributes. Take lessons from Google’s UCP and include expanded content to cover questions about compatibility, availability, policies, and applications. Offer enough depth so that an AI assistant retrieving the content can “explain” your products to users.
  • Funding for protocol operations. Remember that UCP, as well as other protocols and processes, will continue to evolve. Budget for version monitoring, contract testing, certification, and rapid remediation as specifications change.
  • Optimizing for the journey beyond the site. Measure product visibility, referral quality, agent-originated demand, and downstream outcomes when discovery begins inside video, chat, or another company’s interface. Despite a lack of visibility into the conversations that deliver shoppers to your content, there are emerging solutions to optimize for these journeys.
  • Protecting trust at the handoff. Make price, availability, substitutions, returns, identity, and consent explicit when an agent moves from recommendation to transaction. At times, that might mean declining a transaction if it sets off trust alarms.
  • Remaining in learning mode. This area will continue to evolve, and we don’t know all the ways it will yet. Internally, manage expectations for the fact that this is the beginning of a move toward a new way of selling. Before it stabilizes, there will be methods, strategies, and technologies that come and go.

Stay tuned for our upcoming research on the future of owned search in apps and on websites!

Join our webinar for clients and prospects on October 7, 2026, including an advance preview of the research before it publishes!

Share