Amazon Seller Assistant Plugin Comes to Claude
Amazon is letting a third-party assistant read live seller data and act on it. The always-on part is the bigger change.
Amazon announced on 23 September 2026 that its Seller Assistant now exposes a Selling Partner plugin, launching in Amazon Quick and in beta with Anthropic's Claude. A seller can connect listing data, live performance metrics, inventory levels and sales analytics to an assistant that is not Amazon's own.
That is a notable move for a platform that has historically kept seller tooling inside Seller Central. Amazon's announcement frames it as bringing the assistant's reasoning and ability to act into whichever AI tool the seller already uses.
What shipped
Three things, and they are worth separating because they carry different weight.
**The plugin.** Seller Assistant's data and actions become available inside Amazon Quick and, in beta, inside Claude. GeekWire's coverage frames this as Amazon opening its seller tools to outside agents, which is the accurate reading: the plugin is the interface, Claude is the first external consumer.
**Persistent memory.** The assistant retains a seller's pricing patterns, inventory cycles and stated growth goals across sessions, and that memory carries into Quick and Claude rather than staying in Seller Central.
**Always-on workflows.** Workflows now run continuously in the background, including when the seller is not logged in.
The assistant itself is built on Amazon Bedrock using Anthropic's models, with Amazon stating that seller data stays within Amazon's infrastructure and is not shared outside Amazon. The plugin is in beta for sellers in Amazon's US stores, with international expansion described as following. Amazon is also offering primary account holders a free twelve-month Amazon Quick Plus subscription through 31 December 2026, extendable to two additional colleagues.
The headline is not the change
Being able to ask Claude about your Amazon inventory is convenient. It is not structurally different from any other data connector, and a seller who currently exports a report and pastes it into a chat window gets a cleaner version of something they already do.
The background workflows are the actual shift. A tool that only works while you are looking at it is a tool. A tool that watches inventory and pricing while you sleep is closer to an employee, and it inherits the questions you would ask about one: what is it allowed to change on its own, what does it escalate, and how do you find out what it did.
Reporting on the release describes merchants approving actions, which is the right default and the detail worth confirming for your own account before enabling anything. The distance between "drafts a price change for approval" and "changes prices" is the entire risk surface, and on a marketplace where a mispriced listing sells out in minutes, it is not a small one.
What a seller should decide first
Four questions, none of which need a technical answer.
**Which actions require your approval?** Find the setting rather than assuming the default. Price and inventory changes are the two where an error is expensive and fast.
**What does the memory hold?** Persistent memory that carries pricing strategy across into a third-party assistant is useful and is also a record of how you price. Know what is stored and whether you can clear it.
**Who else in the account has this?** The free Quick Plus seats extend to colleagues, and an assistant with action permissions is a permission grant, not a productivity perk.
**What is your check on it?** Background automation fails quietly. A weekly look at what changed, compared against what you expected, catches drift that no alert will.
The general version of this question, for any agent you give account access to, is covered in what to do if an AI agent takes an action you did not approve. The narrower money question is in can an AI agent spend your money.
Why it matters beyond Amazon
Platforms have generally responded to AI assistants by building their own. This is a large one deciding that the assistant layer is not where it competes, and that keeping seller data inside its walls while letting outside agents reason over it is the better trade. If that judgement proves right, other marketplaces will follow, and the practical shape of running a small online business becomes less about which dashboard you log into and more about which assistant holds your context.
For a seller today the sensible posture is unexciting: enable the read side, leave the write side behind an approval step, and revisit in a month once the beta has some mileage. The broader case for where AI genuinely helps a small operation, and where it mostly adds supervision work, is in AI for small business: what actually works.
How did this land?
About the author
Senior Editor, AI & Product
Cecilia leads the Swarmz editorial desk. She has spent a decade turning complex AI and product topics into writing people actually finish, and she owns the blog's quality bar.
