AI Product Recommendations in Helpdesk

Problem Statement

In conversations with pre-sales intent (for example, product discovery or recommendations), agents currently need to:

This results in:

To increase adoption of AI features in the helpdesk by enabling contextual AI product recommendations.

This aims to:

When a customer message with pre-sales intent is detected (for example:

“Can you suggest a shirt for summer?”)

The AI automatically generates a set of relevant product recommendations from the merchant’s catalog.

AI detects pre-sales intent in the incoming message
AI selects relevant products based on:
- Chat context
- Product category and relevance
Recommended products appear as a pre-filled recommendation block inside the chat editor
Products are displayed as product cards
The agent’s workflow is optimised as follows:
- Primary action: Review and send
- Secondary actions:
  - Add more products
  - Remove AI-suggested products
  - Edit the accompanying message

Store admins can control visibility using the Helpdesk AI Assistant toggle:

The feature triggers only when pre-sales or high buying intent is detected in an agent-assigned chat.

Example message:

“Can you suggest a shirt for summer?”

Customer sends a message with pre-sales or buying intent
Chat is assigned to an agent
AI detects high buying intent in the conversation
A Product Recommendation button appears in the chat
Agent clicks the button
AI generates:
- A set of relevant product recommendations (up to 10)
- An AI-drafted message based on the chat context
Products appear as product cards inside a pre-filled recommendation block
Agent can:
- Review the message
- Edit the text
- Add more products
- Remove AI-suggested products
Agent sends the final message to the customer

Single Product Recommendation
- If only one product is shared, it is displayed as a single product card in the chat.
Multiple Product Recommendations
- If multiple products are shared, they are displayed as a product carousel.
- Rearrangement of products is possible inside product picker.

Agents have full flexibility to:

This ensures brand tone, accuracy, and human judgement are always preserved.

Can store admins turn this feature on or off?
Yes. Once enabled, admins can control visibility using the Helpdesk AI Assistant toggle in the dashboard.
Can agents manually trigger product recommendations?
No. Recommendations are triggered automatically based on pre-sales or high buying intent detection.
Does AI automatically send product recommendations to customers?
No. Agents must always review, edit (if needed), and manually send the message.
Is any AI credit deducted when product recommendations are generated?
No. AI Product Recommendation does not consume any AI credits.
Why doesn’t the product recommendation button appear in some chats?
Possible reasons include:
-Feature not enabled for the store
-Helpdesk AI Assistant toggle is OFF
-No pre-sales or buying intent detected
-Recommendation already shown once in the session
How are products displayed to customers?
A single product is shown as a single product card
Multiple products are shown as a product carousel
Can attachments be sent along with product recommendations?
No. Attachments and product recommendations cannot be shared together.
Which channels support this feature?
WhatsApp and Instagram only.

For further assistance or to raise feature requests related to AI Product Recommendation, please contact [email protected]