Most Amazon sellers have already tried using ChatGPT or Claude for their business. The workflow usually looks like this: download a report from Seller Central, paste it into the chat, get a sensible answer that is already out of date, and repeat next week. If you want to connect ChatGPT or Claude to Amazon Seller Central properly, the data problem is only half of it. This guide covers how the connection works technically, how to keep it safe, and the method that makes it useful after day two.

The real problem: your AI can't see your seller account

Large language models are excellent at reasoning, but out of the box they know nothing about your catalog, your prices, your FBA inventory or your Amazon Ads campaigns. They only know what you paste in. That creates three hard limits:

So the AI ends up as a smart consultant working in the dark.

How to connect ChatGPT or Claude to Amazon: MCP and official APIs

The clean way to give an AI access to external systems is MCP (Model Context Protocol), the open standard that ChatGPT and Claude use to talk to tools. An MCP server exposes a set of tools such as "read orders", "read FBA inventory", "read campaigns" or "prepare a price change". The model picks the right tool based on your question, calls it, and reasons over live data.

Vendooly is an MCP connector built exactly for this. It links ChatGPT and Claude to your seller account through official Amazon APIs: the Selling Partner API (SP-API) for Seller Central and the Amazon Ads API. No scraping, no shared passwords. You authorize access through Amazon's standard consent flow and can revoke it at any time. The setup steps are in Connect Seller Central, and there are dedicated walkthroughs for ChatGPT and Claude.

Think of it as an action layer for Amazon, not a dashboard and not another AI. It works with the assistant you already use. Today it exposes 106 tools: 65 read tools and 41 action tools, with 44 dedicated to Amazon Ads (Sponsored Products, Sponsored Brands, Sponsored Display) and 7 for Vendor Central. Coverage spans catalog and listings, pricing and Buy Box, inventory and FBA, orders and sales, account health, catalog quality audits and sales-drop diagnosis.

The safety rule: the AI prepares, you approve

The first question every seller asks is: "What if the AI changes a price by mistake?" Fair question. That's why every write operation (prices, stock, listings, bids, budgets, campaign status) follows the same pattern:

This is deliberately not autopilot. The heavy lifting (finding, cross-referencing, preparing) is done by the model; the decision stays with you. The details are in Action approvals.

Why technology alone isn't enough

Here's what we learned from onboarding sellers and agencies. Once the seller account is connected, the first reaction is excitement. Two days later the question becomes: "OK, so what do I actually ask it?" When a tool can do almost anything, the blank prompt box is the real bottleneck.

So on top of the connector we built a method in three parts.

1. A library of ready-made prompts

Inside the app there are 27 ready-made prompts in 5 languages, grouped into 7 areas: seller account overview, catalog and listings, pricing and Buy Box, inventory and FBA, advertising, GPSR/EPR, and orders and sales. 15 are analysis prompts (read-only) and 12 are operational: they prepare the change, show the preview and wait for your approval.

2. The Monday recap

Every Monday, if your seller account is connected and you had sales in the last 14 days, you get an email with: last 7 days vs the previous 7 (revenue, units, %), Buy Boxes won and lost, products that went out of stock while selling, keywords that spent without selling (if Amazon Ads is connected), your top 2 growing and top 2 declining SKUs, ad spend and ACoS vs last week, and a section called "what to ask your AI this week" with prompts chosen from your own data.

3. A public wiki

The Vendooly wiki has 41 guides across 16 sections in 5 languages, from getting started to operational recipes like the Monday routine.

Your first three prompts after connecting

If you've just connected your seller account, start here. Copy and paste:

Give me an overview of my seller account on amazon.co.uk for the last 30 days: revenue, units, week-by-week trend and the 5 SKUs that matter most. End with three things I should check today.
List the products that sold in the last 30 days but now have low or zero FBA stock. For each one, tell me whether it has active ad campaigns.
Which ASINs lost the Buy Box in the last 7 days? Show my price next to the Buy Box price. Don't change anything, I just want the picture.

The last line of the third prompt is a good habit: when you only want analysis, say so. Nothing is written without your confirmation anyway, but explicit intent keeps the conversation focused.

What to do this week

If you're already copy-pasting reports into ChatGPT or Claude, try this for seven days: every morning, ask your seller account one question instead of opening three reports. Track how long it takes. Our beta testers report saving 5-10 hours a week on operations. That time doesn't come from a magic feature; it comes from no longer searching, exporting and cross-referencing by hand.

Want to try it on your own seller account? The 14-day free trial includes the full Seller + Ads plan, and below €2,000 of Amazon revenue over 30 days it stays free. Sign up at vendooly.com/signup.html, or start with the wiki to see how it works step by step.