Sound familiar?

  • AI assistants describe your products from old or scraped data, or don't mention you at all.
  • The prices and stock an agent sees don't match your shop.
  • Nobody has decided how an AI agent should be allowed to pay, or how you'd know it's genuine.
  • Half the inbox is "where's my order?"
  • The same checkout problem is reported in 40 tickets before engineering hears about it.

Selling through AI agents

When a shopper's assistant goes looking, your products should be easy to find, accurate and ready to buy.

  1. Catalogue agent

    Cleans up and fills in your product data, such as titles, sizes, materials and compatibility, so assistants can match your products to what shoppers ask for.

  2. Live stock and prices

    Publishes your catalogue, stock levels and prices in the formats AI shopping agents read, and updates them as they change, so nobody sells what you don't have.

  3. Agent checkout

    Lets AI agents buy through your existing checkout and payment provider. Payments are tokenised, so card details never pass through the agent.

  4. Rules for agent buyers

    You decide which agents can buy, set limits and choose what needs the shopper's confirmation. Every agent order records who authorised it.

Looking after your customers

  1. Phone agent

    Answers order, delivery and returns calls using live order data, and passes complaints to your team with a summary.

  2. Email agent

    Drafts replies to "where's my order?", returns and product questions from your order and stock data, for your team to approve.

  3. Refunds agent

    Prepares refunds and credits within your policy for one-click approval, and passes exceptions to a person.

  4. Incident agent

    Spots patterns across tickets, such as a broken promo code or failing payments, and raises one issue with engineering. When it's fixed, every affected customer gets a reply.

  5. Notes agent

    Summarises every call and chat on the order and customer, so the next person has the full story.

  6. Review agent

    Checks replies against your tone, your returns policy and consumer law, and shows you where they fall short.

In practice

A shopper asks their assistant for waterproof walking boots in a size 9, under £150, delivered by Saturday. The assistant looks for shops whose product data says “waterproof”, whose stock shows a size 9 and whose delivery promise fits.

With Ordool, your catalogue agent has already filled in those details, and your stock and prices are published as they change. The assistant finds your boots, the shopper confirms, and the order goes through your normal checkout with a tokenised payment. If the shopper later asks where the boots are, the email agent answers from the same order record.

When something goes wrong, such as a discount code that stops working on a Friday evening, the incident agent notices the pattern from the first few tickets and raises a single issue with engineering. When it’s fixed, the email agent drafts a reply to everyone affected, and your team approves it once.

The agents work with the tools you already use, such as Shopify, Zendesk, Gorgias and Stripe. They never see full card numbers.