EN
Connect your AI agent

Add Hello Retail to your AI assistant

For Hello Retail customers: Copy the address and add it as a connector in Claude, ChatGPT or Codex.

MCP address

https://core.helloretail.com/mcp

Not a customer yet?The MCP needs a Hello Retail login. Tell us what you want to build.

Let's talk
Let's talk

Build your own ecommerce personalization

What changed when AI tools learned to build stores, and what it means for the person who runs one

Published

Building your own ecommerce personalization means your store runs on your ideas, carried out by your AI tool on data and products that already work.

Until recently, an idea for the store had to pass through someone else before it reached a shopper: A developer, an agency, a support ticket, or an afternoon in a settings screen. AI coding tools removed most of that distance. A store owner can now describe what they want and have it built or changed in the same conversation.

This guide explains what that shift changes for search, recommendations, pages, email and sponsored placements, what an AI tool cannot bring on its own, and how the pieces fit together at Hello Retail.

What changed

Tools such as Claude, ChatGPT and OpenAI Codex can write a storefront, a landing page or a checkout flow from a description. That is the new reality of ecommerce, and it is good news for anyone with more ideas than development hours. The person who knows the customers can now be the person who makes the change.

It also moves the bottleneck. When building is fast, the question that decides how good a store is becomes a different one: What is the tool building on? A recommendation box is easy to draw. Knowing which four products belong in it for this shopper, this week, is the hard part, and no amount of code generation produces that knowledge.

An AI coding tool can build a store in a weekend. Now give it over a decade of global buying patterns data and see your business flourish.

What your AI tool is missing on its own

Buying patterns

What people buy together, what they come back for, and what sells in which season. This is learned from behavior over time, and a new build starts with none of it.

Engines that already work

Search that understands a misspelled query, recommendations that adapt to each visitor, category pages that sort themselves. These are products with years of production use behind them, ready to be configured.

Your own store's history

Which searches found nothing last month, which recommendation box underperforms, which newsletter products actually sold. A tool can only improve what it can see.

A safe way to make changes

Drafts, a confirmation step before anything is changed, and a log of who changed what. Speed is only useful when a mistake is easy to catch and undo.

The three parts: Your idea, your AI tool, our data and products

Your idea

You know the store. You know that the rain jackets should lead the category page this week, that a brand campaign has ended, or that shoppers keep searching for something you sell under another name. The idea is the part nobody else can supply.

Your AI tool

You describe the idea in your own words to the tool you already use. It works out which steps are needed, shows you what it found and what it proposes, and asks before it changes anything.

Hello Retail's data and products

The tool connects to Hello Retail through MCP, an open standard for giving AI tools a set of actions they can take. Through that connection it can read your store's data and operate Search, Recommendations, Pages, Newsletter Content, Product Agents and Retail Media. Our guide to what MCP is explains the connection without the developer vocabulary.

What a week looks like

The work itself is familiar. What is different is how long each piece takes and who can do it. A request that used to wait for a free afternoon is asked on Monday morning and reviewed before lunch. These are requests the Hello Retail connection documents an answer for today:

  • “Which searches find nothing, and can you fix them?”Looks, then changes after your yes
  • “When people search for X, show them this product first”Looks, then changes after your yes
  • “Our category page converts badly - why?”Looks, then changes after your yes
  • “Show these products first on our sale page”Looks, then changes after your yes
  • “What should we promote this weekend?”Looks only

Each one starts with the tool looking at your data. Where a change follows, the tool describes it and waits for a yes. Our guide to what you can build, product by product goes through every area.

The outcome is the same one personalization has always been bought for: Sell more, keep customers, and get your week back. The third part is the one that changes most.

What stays the same

  • You decide. The tool proposes and you approve. Most changes land in a draft and are published in My Hello Retail.
  • The storefront still runs on code. The search box and the recommendation slots your shoppers touch are built against the API, so they behave the same way on every visit. Our guide to MCP and APIs covers where that line sits.
  • Clean product data matters as much as before. A capable tool working on a badly described catalog gives quick answers about a store that is still hard to shop.
  • Results are measured the same way. Revenue through search, recommendations and email is reported as it always was, and the tool can read those reports for you.

Where to start

Connect the tool you already use, confirm it sees the right store, and ask one question you would normally open a report to answer. Then make one small change and read the confirmation before you approve it. Our guide to connecting your AI tool to your store walks through those first steps.

After the first week the useful question changes from how the connection works to what to do with it. What will you build next?

See the connection at work on the Hello Retail MCP page.

Frequently asked questions

What does it mean to build your own ecommerce personalization?

It means the ideas for how your store should behave come from you, and you carry them out yourself by describing them to the AI tool you already use. The tool works on Hello Retail's data and products (Search, Recommendations, Pages, Newsletter Content, Product Agents and Retail Media), so you are building on engines that already run in production. You are not writing a search engine or a recommendation model from nothing.

Do I need a developer to do this?

Not for the daily work. Someone usually helps with the first connection, since it needs a Company ID and a sign-in. After that the requests are written in ordinary language: Which searches find nothing, show these products first on the sale page, set up a campaign page for summer. A developer is still the right person for anything built into the storefront itself.

Can an AI coding tool build personalization without a platform behind it?

It can build the interface: A search box, a carousel, a page. What it cannot produce on its own is the knowledge of what people buy together, what sells in which season, and which result a shopper meant. That knowledge comes from buying patterns collected over time and across many stores. The practical answer is to let the tool build and operate, and to give it that data to work on.

Does my AI tool change the live store on its own?

No. It asks before it changes anything, and most changes land in a draft that you publish in My Hello Retail. A small number of changes take effect when they are saved, and every tool is labeled read-only, write or destructive write. Our guide to what your AI tool can change lists the exceptions.

Which AI tools does this work with?

Hello Retail documents setup for Claude, ChatGPT and OpenAI Codex. The connection uses the Model Context Protocol, an open standard, so the list is set by which tools support that standard and not by a private integration.

Is the dashboard going away?

No. My Hello Retail is where drafts are published and where everything can still be set up by hand. What changes is the default route for an idea: You describe it, read what the tool proposes, and approve it.

Build the store no one else has

Connect Claude, ChatGPT or OpenAI Codex to Hello Retail and start with one request. Setup steps for each tool are in the developer documentation.

Start building with our MCP