Several weeks ago, Amazon announced Nova Act, an AI model that can search for products, add to cart, and make purchases on someone’s behalf without intervention. As AI agents like Nova Act become more popular, marketers will need to target not only people, but the machines operating on their behalf, using a strategy called machine-to-machine—or M2M—marketing.
While M2M marketing might be a buzzword today, agencies like VML and Digitas and tech companies like Microsoft are preparing for a future driven by M2M marketing. Here’s what marketers need to know:
“Machine-to-machine marketing is exactly what it sounds like,” said Jason Carmel, global creative data lead at VML. “It's literally two machines, kind of navigating communication with each other to get the best for each of their particular beneficiaries.”
M2M marketing occurs when two AI agents—one for the brand and one for the consumer—navigate decision-making. The term indicates a future where brands optimize for AI agents’ choices, rather than people’s decisions.
“The connection between humans and brands is still at the center, with machines facilitating communication and translating needs back and forth between the two,” said Jen Faraci, chief data officer at Digitas.
“Agentic AI is a part of machine-to-machine marketing,” said Carmel. “If you have two agents talking to each other and they're both machines, congratulations, you have machine-to-machine marketing.”
Consumers can chat to these AI agents. Amazon’s Rufus can give recommendations to shoppers. Shopfiy’s Shop AI can help consumers discover new products. Expedia’s Romie can help travelers plan trips. Each of these AI agents interact with real people, meaning they’re not currently using M2M marketing.
“Most agents currently interact with humans,” Faraci said. “However, we are starting to see the beginnings of agents interacting with other agents. This evolution marks a significant shift in how machine-based systems operate autonomously.”
M2M marketing hasn’t yet reached consumers.
“Marketers are using machine-to-machine marketing quite a bit,” Carmel said, referring to “the ability for a marketer to not have to tell a machine what it needs.” Programmatic advertising could be viewed as a form of M2M marketing because technology is making targeting decisions on behalf of individuals. But machines still aren’t making choices on behalf of consumers.
“When it comes to creating meaningful human-to-brand connections, true machine-to-machine marketing isn’t quite there yet,” Faraci said.
“We anticipate that brands will soon have their own AI agents, which are essentially software programs that can quickly perform tasks and make decisions,” wrote Kya Sainsbury-Carter, corporate vice president of Microsoft Advertising, in a blog post. “For example, in the same way that a travel agent could historically help you find the perfect vacation, a software agent can help you find what you’re seeking to consume or purchase, moving us into the era where brand agents can have conversations with people and even with other companies and their agents.”
AI agents could theoretically make all decisions for a task like booking a vacation, Faraci said. “For those who are willing to give up some control, agentic-based decision-making is on the horizon and progressing rapidly.”
Not yet, and that resistance could make M2M marketing unnecessary. “Technologists may be overestimating how much people want to automate shopping,” wrote our analyst Yoram Wurmser in his “AI Agents and the Consumer Journey” report. In fact, 66% of US consumers would not allow AI tools to make purchases on their behalf, according to February 2025 data from Omnisend.
But there are some instances where the majority of consumers are already interested in AI agents, such as securing high-demand products before they sell out (66%), buying items when they reach a target price (65%), and monitoring purchased products for maintenance (63%), according to December 2024 data from Salesforce.
“I think it will become much more normal for people to be comfortable letting their machines investigate, be marketed to, learn about something, and then make a purchase decision,” Carmel said, indicating a need for M2M marketing. “Five years from now, we will look back on ourselves and have a slightly embarrassed giggle at how shy we were” about allowing AI agents to make purchases.”
This was originally featured in the EMARKETER Daily newsletter. For more marketing insights, statistics, and trends, subscribe here.