Why Mondelez is hiring a global lead to solve for AI-driven shopping bots

The role originally posted in February, and called for someone to help the snacking company shape global strategy, run pilots and build out a scalable playbook.

“We’re at the beginning of a fundamental shift in commerce — moving from clicks and channels to intelligent agents that guide decisions, discovery and transactions,” Andrew Lederman, vp of global digital commerce at Mondelez, wrote on LinkedIn sharing the job post.

Agentic uptick by 2028 

Mondelez’s own retailer partners said they were expecting 30% of their site traffic to be agentic by 2028. By 2030, the global market for agentic commerce will be worth between $3 trillion and $5 trillion, according to McKinsey estimates. In the U.S., retail sales to consumers via agentic commerce could reach up to $1 trillion. 

Those numbers are nothing to scoff at and likely fanning marketers’ sense of urgency over search’s shifting landscape. 

“Everyone right now, very fairly, needs to prioritize [agentic commerce] because it’s a frightening shift from the way they’re used to approaching consumers,” Mike O’Donnell, svp of innovation and business transformation at Flywheel, told Digiday. 

Platforms have already gotten the jump. For example, Google began rolling out the Universal Commerce Protocol (UCP) earlier this year — a foundational layer that aims to enable agentic shopping in the tech giant’s ecosystem. 

Late last year, Amazon’s AI shopping assistant Rufus rolled out an auto-buy feature, allowing the AI agent to automatically purchase items for users when it hits a certain price. 

Around the same time, there was also OpenAI’s instant check out feature in ChatGPT. Shoppers could tell ChatGPT what they wanted and the platform would make recommendations. If a merchant used the Instant Checkout feature, shoppers could confirm their address and pay within the ChatGPT app.  

OpenAI launched the feature in September. Just a few months later, the tech behemoth pivoted after lackluster results. That move speaks to bigger questions around agentic commerce, user adoption, walled gardens and infrastructure. 

Goldrush that feels reminiscent 

In some part, the agentic commerce gold rush feels reminiscent of the metaverse, non-fungible tokens (NFTs), and to some extent, social commerce.

“[Agentic commerce is] like social commerce, which is actually still a huge buzzword with our clients,” said Leah Sallen, managing director of retail and commerce media at VML. She added, “I feel like we haven’t even gotten past that yet. Now it’s social commerce plus agentic, plus AI.”

Given the unknowns, Sallen is directing clients to focus on the macro picture — shoring up generative engine optimization strategies to make product information AI discoverable. 

Flywheel’s O’Donnell is taking a similar approach. At least one client, he said, is sending specific product data to OpenAI through a model context protocol (MCP). Essentially, O’Donnell explained, this allows the AI agent to check against real product data and ultimately ensure that the AI’s response is accurate. 

Google’s UCP is a good start, but brands are still determining how to build their own internal infrastructure — things like setting up MCP servers, which essentially allow LLMs to talk with the outside world.

There’s also the walled gardens. Some shoppers may use retailers’ agents, like Amazon’s Rufus or Walmart’s Sparky, while others turn to standalone platforms, making it difficult for advertisers to scale an agentic commerce playbook.

Finally, there’s the trust factor. While people use AI agents to shop and research, transaction is a different story. OpenAI’s Instant Checkout hiccup is an example of that. 

“Traffic is the biggest focus now,” said O’Donnell. “Can we get consumers to change behavior to start clicking through more of those sponsored prompts to increase the actual value of those conversions?”

As Lederman said on the Digiday Podcast, agentic commerce is less a matter of if it’ll happen and more a matter of when. 

Or as O’Donnell puts it: “There’s realities of today. There’s probably opportunities for tomorrow that are yet to be determined.”