How Adobe is building AI and agentic tools for brands

Ready or not, the age of AI-assisted online shopping is here.

Traffic to US retail websites coming from AI spiked 393% year over year in the first three months of 2026, according to recent data from Adobe.

That’s not all: “What’s also caught the attention of brands is the conversion of that traffic,” Loni Stark, VP of strategy and product at Adobe, told Marketing Brew. “What they found was that traffic coming from these LLMs are higher converting, and they’re worth more than just average traffic.”

With AI increasingly facilitating online shopping, some brands are looking for new ways to scale their presence in AI tools while also figuring out how to effectively deploy agentic AI in-house—and Adobe wants to be part of that scaling effort. The software company is building out brand visibility and agentic AI tools that are aimed at helping brands stand out to LLMs while allowing them to utilize agentic AI tools across their enterprises, according to Stark.

When it comes to agentic AI in particular, Adobe is keeping an eye on what aspects of agentic protocol, which enable agent communication, actually get adopted.

“The standards are there,” Stark said. “Adobe Commerce and the rest of the Adobe portfolio is committed to supporting them, but we are doing [it in] a very pragmatic way with our customers to see which parts of [agentic AI] they’re adopting first.”

Pick me, choose me, love me: Buying through retailers’ storefronts, websites, and mobile apps remains dominant with consumers, Stark said, but brands are increasingly investing in optimizing their AI presence and integrating AI into their digital touchpoints. To convert AI searches into sales, it helps for a brand to be noticed by the AI first.

At its Adobe Summit event in Las Vegas earlier this month, Adobe announced several new tools designed to boost brand visibility and consumer engagement to help with that noticing. That includes an enhanced Brand Concierge tool, which uses AI agents to build chat features that consumers can interact with across a brand’s sites and apps, along with real-time product information and checkout options.

Adobe is also putting out LLM Apps, a feature in its content and asset management platform Adobe Experience Manager that is designed to help brands build “branded experiences that run directly within LLM interfaces,” according to a company press release. For example, a user looking for the running shoes who has installed the Dick’s Sporting Goods app within ChatGPT can get product recommendations there, Stark said, by way of example. The tool could query users further to get information on sizing or other specifications, and it could also drive consumers to Dick’s Sporting Goods’s website, she added.

“That’s a connection where you’re starting a conversation, you’re getting to a deeper brand experience in LLM Apps, and then being able to click through to the richer experience that brands are able to offer on their own property,” Stark said.

Go with the flow: While optimizing their AI presence, brands are also increasingly thinking about leveraging agentic AI. To that end, Adobe rolled out its agentic AI system CX Enterprise aimed at creating composability, or the ability for different technologies to connect or interact with one another, so brands can upgrade their own agents with Adobe offerings, Stark said. That could look like brands using CX Enterprise’s agent skills catalog, which equips agents to take actions, including planning campaigns or building relevant audiences, with their own agents.

As brands get more sophisticated with agents, it’s likely they’ll need to orchestrate multiple agents, aka coordinate them within one system, and to support that need, Adobe is building partnerships with AI companies like OpenAI, Microsoft, and Anthropic, which run their own agentic offerings.

But Adobe isn’t all-in on every aspect of AI. Some agentic AI protocols are seeing more adoption than others, according to Stark, and Adobe is focused on the ones with the highest use. The company has seen strong adoption of the Model Context Protocol, which allows agents to connect to other tools, Stark said, and Adobe has subsequently built out related offerings for brands to manage MCP.

However, adoption of the Agent Communication Protocol, which supports agent-to-agent interactions, has been lower, Stark said, and when it comes to Google’s more e-commerce-focused Universal Commerce Protocol, Adobe has so far adopted the parts of it focused on product discovery.

Of course, as consumers get more comfortable with different tools and protocols, that all stands to change.

“As we see more of that adoption in working with our clients [in terms of] the transaction pieces of it, we have [more parts of UCP] on our roadmap,” Stark said.