Amazon and Walmart continue to duke it out for dominance in the commerce realm. They represent fundamentally different approaches in online versus offline shopping, while their e-commerce platforms are increasingly competitive. Now the latest divergence is evident in their strategies: agentic shopping.
Specifically, Amazon is resistant, while Walmart appears to be all in. The former can be seen in Amazon’s well-publicized moves to block AI engines from its domain. This includes those that scrape its content for AI search, as well as those that act on behalf of users to shop and transact, such as Muse.
Meanwhile, Walmart’s embrace of agentic shopping is underscored by several moves. For example, it has formed partnerships with Muse, ChatGPT, and Gemini to give them special access to its eCommerce pages and systems so their agents can do their thing (see our Muse coverage yesterday).
Stepping back for context, what are we talking about here? Agentic commerce involves agents that shop autonomously. Driven by prompts, they find and purchase products. The appeal lies in two areas: saving time and elevating capabilities. The latter includes capturing limited-window deals or finding rare items.
Agentic Embrace
Back to Walmart, the list of moves that signal its agentic embrace is growing. For example its new Agent Gateway is an MCP server designed to help agents connect to Walmart and gain the access that they need to shop. This takes the partnerships noted above and scales them up to a long tail of others.
The company is also developing customer-facing features built for the agentic shopping age. For example, it will differentiate between customer’s manual purchases and agentic ones. So when looking at one’s order history, users can see both lists, and adjust settings for what they allow agents to do.
Taking that a step further, next month Walmart will roll out what it calls “agent profiles.” This is sort of like a Netflix profile where users can create several shopping personas that have their own functions, purchase history and capabilities. For example, each can be tuned to spending limits or permissions.
Beyond Netflix profiles (our analogy) Walmart positions this as sort of like adding a family member to your credit card account – they get their own credit card number that you can manage, and other controls like spending caps. This framing aligns with an ongoing mindset that treats agents like people.
Demarcation Line
When you add up all the above – personifying agents, setting parameters, etc – it’s all about giving users greater trust and confidence in agentic shopping. This is what the emerging area needs at this point: AI is inherently plagued by trust issues, which amplifies in the context of autonomous transactions.
Moreover, we’ll continue to see divergence in approaches around agentic shopping. Given definitive moves from Walmart and Amazon, the two biggest players in eCommerce – at least in the U.S. – have chosen their sides. Several tertiary players will likewise stand along the agentic commerce dividing line.
What do those camps look like so far? Joining Amazon in blocking agentic shopping are players such as eBay, Pizza Hut, Adidas, and several airlines. And on Walmart’s pro-agentic side, the most notable player to pledge its allegiance is Shopify – recently enabling agentic shopping access across its platform.
The ultimate question is if this demarcation line is in permanent ink. We tend to believe that anyone rejecting agentic commerce, including Amazon, will eventually come around. Once the user trust hits a tipping point, anyone who isn’t on the agentic commerce train will be leaving revenue on the table.
