Shopify AI Product Recommendations on WhatsApp | ZazzyAgent

Customers don't always know the exact product they want.

They might say:

I need running shoes for daily use.

or:

I'm looking for a laptop under ₹70,000.

or:

Can you recommend something for a wedding?

An AI Agent can understand the requirement and use your connected Shopify store to find relevant products.

What the agent can do

A Shopify sales agent can:

• Understand customer requirements

• Ask follow-up questions

• Search relevant products

• Recommend suitable products

• Continue the conversation

• Save customer preferences

• Hand the conversation to sales when appropriate

Before you start

You need:

• Connected WhatsApp

• Connected Shopify store

• AI Agent

• Product information

• Shopify product recommendation action

If you haven't connected Shopify yet, start with ZazzyAgent Integrations & Developer Guides.

The Shopify store action

Shopify actions require a store to be selected first.

Use:

##shopify_store## : 12

Select the connected store using the ## autocomplete.

The product recommendation action

Use:

##shopify_recommend_products##

This lets the agent search and recommend products from the connected Shopify store.

Example System Prompt

You are the Sales Agent for our Shopify store.

Help customers discover products that match their needs.

Ask short follow-up questions when important information
is missing, such as:

- Product type
- Budget
- Size
- Color
- Intended use

When the customer asks for a product recommendation,
search the connected Shopify store.

##shopify_recommend_products##

Recommend only products returned by the connected store.

Do not invent products, prices, stock information,
discounts or product specifications.

Keep recommendations concise and explain why each product
matches the customer's requirements.

Don't recommend before understanding the customer

Customer:

I need shoes.

That's incomplete.

The agent could ask:

Sure. What will you mainly use them for?

Then:

What's your approximate budget?

Then search for appropriate products.

Ask only useful questions

Don't turn every product recommendation into a long questionnaire.

Ask what actually changes the recommendation.

For shoes:

• Type

• Use

• Size

• Budget

For electronics:

• Use case

• Budget

• Required features

Example conversation

Customer:

I need running shoes.

Agent:

Sure. What's your approximate budget?

Customer:

Around ₹5,000.

Agent:

Do you need them mainly for daily running or occasional use?

Customer:

Daily running.

Now the agent has useful context.

It can call:

##shopify_recommend_products##

and recommend relevant products.

Don't invent stock

Never instruct the AI to say:

This product is definitely in stock.

unless the connected store/action provides that information.

The System Prompt should say:

Use the returned Shopify information. Never invent live product availability.

Don't invent discounts

If the store returns:

₹4,999

don't let the AI say:

We can give you 20% off.

unless that discount actually exists in the configured information.

Product recommendation vs product knowledge

These work together.

Knowledge

Explains:

What makes your product different?

Shopify action

Finds:

Which products should I recommend right now?

Save customer preference

You can combine product recommendations with custom fields.

For example:

When the customer clearly identifies their preferred product category,
save it to the Product Interest custom field.

##save_custom_field## : 42

See How AI Agents Save Customer Information in ZazzyAgent.

You can also add a label:

##add_label## : 12

See How to Add and Remove Customer Labels With ZazzyAgent AI Agents.

Transfer serious buyers to sales

If you have a human sales team, you can route high-intent customers to them.

Example:

When the customer clearly indicates that they want to purchase
and requires human sales assistance,
assign the conversation to the Sales team.

##assign_human_role## : 3

Complete sales flow

Customer
   ↓
AI understands requirement
   ↓
Ask useful questions
   ↓
Shopify product search
   ↓
Recommend products
   ↓
Customer interested?
   ↓
Yes
   ↓
Add label / save information
   ↓
Human sales when needed

Common problems

No products are recommended

Check:

• Shopify store selected.

• Shopify action configured.

• Connected store contains products.

• Customer has provided enough information.

AI recommends products that don't exist

Tell it:

Only recommend products returned by the Shopify action.

AI invents price

Add:

Never invent prices. Use only returned store data.

AI recommends irrelevant products

Make the prompt require the agent to ask for missing information before searching.

Don't overload the sales agent

Give it only the Shopify capabilities it needs.

A focused sales agent might need:

##shopify_store## : 12
##shopify_recommend_products##
##save_custom_field## : 42
##add_label## : 12

You may not need order cancellation, order updates or other support actions.

That keeps the agent easier to test and control.