Amazon announced Wednesday the most significant upgrade yet to Seller Assistant, its AI companion for the millions of merchants on its marketplace: the assistant now carries persistent memory of each seller's business, can execute routine tasks around the clock, and — for the first time — extends into AI tools Amazon doesn't own, through a new plugin launching with Amazon Quick and in beta with Anthropic's Claude.

The announcement, detailed in a post by Mary Beth Westmoreland, Amazon's vice president of Worldwide Selling Partner Experience, marks a notable shift for a company that has historically kept seller data and workflows inside its own walls. For more on how the AI industry's biggest players are competing for the enterprise, see our latest AI developments.

An Assistant That Remembers

The headline capability is memory. Seller Assistant now maintains persistent knowledge of each seller's pricing patterns, inventory cycles and growth goals, and that context follows the seller across every surface where the assistant lives — Seller Central, Amazon Quick, or Claude. Amazon says recommendations sharpen with every interaction, and each suggestion arrives with the reasoning behind it so sellers can evaluate or override it.

The upgrade also brings a canvas feature, a dynamic visual workspace where sellers can model scenarios, compare pricing strategies and diagnose campaign performance using live data through natural conversation.

The assistant itself is powered by Anthropic's Claude models on Amazon Bedrock, combined, Amazon says, with 25 years of its own commerce expertise. Amazon is explicit about the data boundary: seller data stays within Amazon's infrastructure and is never shared outside the company — a framing clearly designed to preempt the trust questions that arise when an e-commerce giant's data meets an external AI model.

Workflows That Run While You Sleep

The second major piece is automation. Previously, Seller Assistant acted only when prompted — a seller could ask it to adjust a price or draft a restock plan, and it would reason through the task and execute with permission. The new workflows flip that model: sellers describe what they need in plain language, set guardrails, and the workflow monitors conditions and acts continuously, even when the seller isn't logged in.

Amazon's example is telling: "Alert me if any of my top 10 products drop below a 4-star rating and draft a response plan." The company says every workflow runs with full audit trails and seller-defined guardrails — a governance layer that will matter to sellers wary of ceding pricing and inventory decisions to an algorithm.

The Claude Plugin Is the Real Story

The most strategically interesting piece is the plugin. It carries Seller Assistant's intelligence — and sellers' own data — into the AI tools they already use, starting with Amazon's own Quick assistant and extending, in beta, to Anthropic's Claude. Amazon is sweetening the launch with a free 12-month Quick Plus subscription.

That is a small product decision with larger implications. Amazon is effectively acknowledging that sellers live in multiple AI environments, and that an assistant confined to Seller Central will lose to one that meets sellers inside Claude. It also deepens the Amazon–Anthropic relationship, which has been one of the defining corporate partnerships of this AI cycle: Amazon builds on Claude models through Bedrock, and now distributes Anthropic's assistant to its merchant base.

For sellers, the practical appeal is continuity — the same memory and context available whether they are working in Seller Central, Quick or Claude, so nothing is lost switching between tools.

The Numbers Behind the Push

Amazon launched Seller Assistant in 2023 as a generative AI question-answering companion and later added agentic capabilities that let it reason, plan and act on a seller's behalf with permission. The company says the assistant has now rolled out to more than 90% of its selling partners worldwide in their native languages, counts hundreds of thousands of active users, and sees sellers accept its recommendations more than 90% of the time.

Those adoption figures explain why Amazon is pushing further. Sellers, the company concluded, "don't just want AI that works with them. They want AI that works for them."

What It Means

The release is another marker in the shift from chatbots to agents in commerce infrastructure. Competitors are racing to similar territory, and the plugin model — meeting users inside third-party AI tools rather than demanding they visit your dashboard — may prove the template for how enterprise software adapts to a world where the interface is an AI assistant.

For the millions of businesses that sell on Amazon, the near-term reality is more concrete: restocking decisions, pricing adjustments and account health monitoring that used to require a human at the keyboard can now run themselves, with an audit trail to check the machine's work.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →