---
title: "How to deploy an AI product recommendation chatbot: 5 steps"
slug: "ai-product-recommendation-chatbot"
url: "https://chatty.net/blog/ai-product-recommendation-chatbot/"
kind: "post"
author: "Drake Q."
author_url: "https://chatty.net/author/drake/"
published_at: "2026-02-10T09:02:57+07:00"
updated_at: "2026-11-13T10:00:00+07:00"
description: "An AI product recommendation chatbot asks what the shopper needs and shows products they can add to the cart. Set one up on Shopify in 5 steps."
categories: ["AI chatbot"]
tags: ["Chatbots"]
reading_time: 9
feature_image: "https://cdn.chatty.net/wp-content/uploads/2026/02/161-1.webp"
---

# How to deploy an AI product recommendation chatbot: 5 steps

An AI product recommendation chatbot asks what the shopper needs and shows products they can add to the cart. Set one up on Shopify in 5 steps, with menu paths and plans.

An AI product recommendation chatbot asks shoppers what they need, then shows matching products they can add to the cart from the chat. This guide deploys one on Shopify in five steps with Chatty, with the menu path, plan and limit for each step.

The step most stores rush is the one that decides the result: telling the AI which products to recommend. The chat window takes minutes to set up. The product details and recommendation rules behind it take longer, and they are what shoppers actually see.

Chatty is the [AI sales agent](https://chatty.net/ai-sales-agent/) for Shopify stores. AI chatbot and live chat are two standout features of Chatty. The AI chatbot answers from your own Shopify data, and with live chat your team takes over from the AI and closes the sale.

From February 5 to April 8, 2026, Chatty helped [Just Nutritive](https://chatty.net/case-study/just-nutritive/) sell $150K of hair care across 2,912 orders. Of its 7,699 conversations, 4,900 asked for a product recommendation.

**Key Takeaways**

- **A recommendation chatbot asks, narrows, then shows products to buy**

  It turns a vague need into a short list of in-stock products with an Add to cart button.
- **The product rules matter more than the chat design**

  Most setup time belongs in product descriptions and Smart recommendations, because the AI can only recommend what it can read.
- **Best sellers and new arrivals fill themselves**

  Chatty syncs the top 20 of each from Shopify daily. Seasonal and gift picks need Custom recommendations on Pro or Plus.
- **Recommend before the shopper asks**

  Proactive product recommendations on category and product pages reach shoppers who never open the chat.
- **Measure the orders, not the chats**

  The Chatty Direct field marks orders that contain a product Chatty showed or mentioned in the chat.

## What does an AI product recommendation chatbot do?

An AI product recommendation chatbot turns a vague need into one or two products the shopper can buy. It asks a question or two and filters your catalog by budget, size or use. Then it checks stock and shows product cards with an Add to cart button.

Static blocks such as "frequently bought together" work when the shopper already knows the product. A chat works when the need is still vague, such as "a gift for a cyclist under $50."

Shoppers ask this kind of question often. Chatty looked at 1,763 of its stores with 50+ messages in Q3 2026. In the median store, 46% of shopper questions were about what to buy.

Four actions separate a recommendation assistant from a search box:

- **It asks back.** When a request is vague, it asks about size, budget or use before it suggests anything.
- **It checks stock.** It reads stock from your synced Shopify catalog, skips items that are sold out and offers what is in stock instead.
- **It compares.** In Chatty's AI Mode, shoppers select several products and tap Compare to see price, material and fit side by side.
- **It acts in the cart.** Chatty adds the product to the shopper's real Shopify cart, and checkout stays with Shopify.

For the wider category, our guide to [what an AI sales agent is](https://chatty.net/blog/what-is-an-ai-sales-agent-for-shopify/) covers the actions before and after the recommendation.

## How to deploy an AI product recommendation chatbot on Shopify in 5 steps

Deploying a product recommendation chatbot on Shopify takes five steps in Chatty. You show the chatbox, sync the catalog, set the rules, add sizing and proactive recommendations, then test and measure. Each step below names the menu path, the plan and the limit, checked against Chatty's help center in October 2026.

### Step 1. Install Chatty and show the chatbox in AI Mode

Install Chatty from the [Shopify App Store](https://apps.shopify.com/chatty) and finish the setup flow. On the dashboard, find the Embed app to your theme card and click Turn on. In the theme editor, switch on Chatty - Core and click Save.

![Shopify theme editor App embeds panel with the Chatty app embed switched on](https://cdn.chatty.net/wp-content/uploads/2026/02/chatty-enable-storefront-chatbox-1.webp)

Next, go to Chatbox and turn on Display chatbox on your store. In the Chatbox mode card, select AI Mode.

AI Mode puts a search bar on your storefront. Shoppers search, ask the AI and add products to the cart in one panel.

Plan: every plan, including Free. Limit: your store runs one chatbox mode at a time, so AI Mode replaces the Legacy chat bubble from the next page load.

### Step 2. Sync your catalog and write the details shoppers ask about

Chatty syncs your Shopify products on its own: titles, descriptions, variants, prices, inventory, collections and product metafields. Check what it learned in AI agent, Training data, Products. The product data refreshes daily.

![Chatty AI training data screen showing product sync from Shopify and the manage metafields option](https://cdn.chatty.net/wp-content/uploads/2026/02/chatty-sync-products-for-ai-training-1.webp)

The AI can only recommend on attributes it can read. Put advice such as "best for wide feet" in the product description. To keep it off the page, use a product metafield instead.

For one product's extra questions, open the three dots menu on that product and click Manage FAQs.

![Chatty custom knowledge window for adding a question and answer to the AI's training data](https://cdn.chatty.net/wp-content/uploads/2026/02/chatty-training-ai-with-custom-knowledge-1.webp)

Plan limit: the AI learns 200 products on Free, 1,500 on Basic, 8,000 on Pro and unlimited on Plus. Above the limit, the AI uses in-stock products first. A product with no description stays out of AI training, so fill descriptions before you turn the AI on.

### Step 3. Tell the AI which products to recommend

Go to AI agent, Skills. Under Shopping, find Smart recommendations and click the gear icon. This page decides which products the AI suggests when a shopper asks "what's popular?" or "what's new?"

Best sellers and New arrivals fill themselves. Chatty takes the top 20 products in Shopify's best-selling order and your 20 newest products, and refreshes both lists daily. Turn each list on with its Status toggle.

To rank best sellers by units sold in the last 30 days, turn on AI agent, Settings, Best seller ranking. It is off by default and asks for read-only access to your Shopify reports.

On the Shopify recommendations card, turn on Related products, Complementary products or both. The AI then uses the product relationships Shopify already calculates, which is how it suggests the helmet that goes with a bike.

Seasonal picks do not fill themselves. To steer the AI toward gifts, a sale or a seasonal range, create a rule in Custom recommendations, available on Pro and Plus. Each rule takes a title, the search terms shoppers use and the products or collections to suggest.

Plan: Smart recommendations run on every plan, limited on Free and Basic and unlimited on Pro and Plus. Custom recommendations need Pro or Plus.

### Step 4. Add sizing, a product quiz and proactive recommendations

Chatty puts recommendations where shoppers browse, rather than only in the chat bubble. It helps shoppers in the search bar, on product pages, in chat and at the cart.

Three settings extend the recommendations:

- **Size guide (Pro and Plus).** In AI agent, Skills, open Size guide and click Add size guide. Paste or upload your size chart, and the AI asks for the measurements before it recommends a size.
- **Product quiz (Pro and Plus).** In AI agent, Skills, turn on Product quiz. Choose 3, 5 or 7 questions and 1 to 3 recommended products.
- **Proactive recommendations (every plan).** In Proactive chat, click Create proactive chat and choose Product recommendation for category and product pages. Activate it, then click Publish proactive chat.

The Smart Product Page template, also on every plan, adds a floater with a variant picker and in-chat Add to cart on product pages. Shoppers can pick a variant and add it to the cart without opening the chat.

### Step 5. Test the recommendations and measure the orders

Go to AI agent and click Test AI. Ask the questions your shoppers send, in their words. Try "gift for my dad under $50" or "something like this but in blue."

Under each answer, click Review sources to see which product data the AI used. When an answer is wrong, click Add Q&A, or fix the product description and test again.

After launch, check the unresolved questions card in AI agent once a week. It splits them by reason, and Missing knowledge marks the questions your training data could not answer.

Chatty marks every order it assisted in Shopify, so you can check the return yourself. The Chatty Direct field is true when the order contains a product Chatty showed or mentioned in the chat.

Analytics, Sales shows assisted revenue and your top products from chat on any paid plan. The per-topic order numbers need Pro or Plus.

## What AI product recommendations sold for three Chatty stores

What do product recommendations in chat look like on a real catalog? Three Chatty stores show it. Each figure below comes from the store's published case study, with its time frame.

The stores sell very different catalogs: 10,000+ sports items, 400+ western fashion products and a supplement range.

### Decathlon Réunion: a 10,000-product catalog

![Decathlon's first week with Chatty: 2,000+ conversations handled, 9% chat-to-sales conversion and €10,964 in AI-attributed revenue](https://cdn.chatty.net/wp-content/uploads/2026/10/decathlon-chatty-first-week-results.webp)

Chatty learned Decathlon Réunion's catalog of 10,000+ products overnight, with specifications and sizing guides. In its first 7 days, Chatty made €10,964 in assisted revenue and 9% of chats ended in a sale, per the [Decathlon case study](https://chatty.net/case-study/decathlon/).

Product recommendation was the second biggest conversation topic that week, at 409 conversations, all in French.

### Montana West: gifts during the holiday rush

From July to December 2025, Chatty brought [Montana West](https://chatty.net/case-study/montana-west/) $41,115 in assisted revenue, and 11.86% of its 4,240 conversations ended in an order.

The western fashion store had 400+ products. Chatty recommended gifts and matching pieces while the team took the complex style consultations.

### Stonehenge Health: supplements that must be in stock

Over 7 months, Chatty helped [Stonehenge Health](https://chatty.net/case-study/stonehenge-health/) make $75K in assisted revenue across 599 orders. Of its 5,141 conversations, 11.36% ended in a sale.

Chatty trained itself from the Shopify catalog and the store's FAQ pages. It reads stock levels, so it only recommends supplements that are in stock.

## Four mistakes that make product recommendations miss

When recommendations miss, check the data and the rules before you blame the AI model. Here are four common mistakes, and each has a fix in the steps above:

- **Thin product descriptions.** The AI cannot recommend on fit, material or use if the description never mentions them. Fix it in Step 2.
- **No rule for the products you want to sell.** Without a Custom recommendation, a gift question gets whatever matches the words. Fix it in Step 3.
- **Recommending before asking.** A list of ten products for "a gift" helps nobody. Turn on the product quiz or the size guide in Step 4.
- **Counting chats instead of orders.** A busy chat that sells nothing is a support cost. Filter orders by Chatty Direct in Step 5.

## Set the product rules first, then switch on the recommendation chatbot

An AI product recommendation tool can only work with the catalog data and rules you give it. The chat window takes minutes, while descriptions, Smart recommendations and the size guide decide what shoppers are shown.

Do the five steps in order, and judge the result by orders with the Chatty Direct field rather than by chats answered.

To try it on your catalog, start on Chatty's [Free plan](https://chatty.net/pricing/), turn on Best sellers and ask the AI for a gift under $50.

## FAQ

### What is an AI product recommendation chatbot?

An AI product recommendation chatbot asks shoppers what they need, then shows matching products from your catalog as cards they can add to the cart. It reads your product data from Shopify, including stock, so it can skip items that are out of stock.

### How long does it take to set up a product recommendation chatbot on Shopify?

The basic Chatty setup takes about 15 minutes. You finish the setup flow, embed the app, turn on the chatbox and check a live page. Product descriptions and recommendation rules take longer, and they decide the quality.

### Can the chatbot recommend best sellers automatically?

Yes. In Chatty, Best sellers takes the top 20 products in Shopify's best-selling order, New arrivals takes your 20 newest products, and both lists refresh daily. You turn each list on in Smart recommendations and can add or remove products by hand.

### How much does an AI product recommendation chatbot cost?

As of October 28, 2026, Chatty's Free plan includes 50 AI conversations a month. Custom recommendations, the size guide and the product quiz need Pro, at $68.99 a month for 500 AI conversations. Your plan includes AI conversations. Each AI reply over the plan costs $0.075, up to the monthly cap you set.

### Is Chatty a chatbot?

Chatty is an AI sales agent rather than a scripted chatbot. It writes each answer from your store's data instead of following a fixed script, and it can take actions such as adding products to the cart and looking up an order. Your team also gets a live chat inbox for the conversations that need a person.

### Can a product recommendation chatbot work on WhatsApp or Messenger?

Chatty's recommendations are built for your storefront: the search bar, product pages, the chat and the cart. Messages that arrive on WhatsApp, Instagram or Messenger land in one inbox, where your team can reply and send product links.
