Creating a conversational shopping experience through WhatsApp was both a design and technology challenge. The goal was simple: allow users to browse, choose, and purchase products in a natural, intuitive way, just by chatting.

To make this possible, we partnered with Google and Meta to leverage LLM (Large Language Model) technology. Instead of building a rigid chatbot with predefined flows, we focused on creating a system that could understand intent, adapt to different user behaviors, and guide people through the shopping journey conversationally.

One of the main challenges was balancing flexibility and control. While the AI could interpret open-ended messages like “I’m looking for a new TV” or “show me something under $500,” we needed to ensure the responses stayed relevant, accurate, and aligned with Magalu’s catalog.

This led us to design a hybrid experience:

  • AI handles interpretation and conversation

  • Structured systems handle product search, pricing, and checkout

Another key focus was reducing friction. Traditional e-commerce flows rely on multiple screens, filters, and clicks. In this experience, we translated those steps into a dialogue:

  • Asking preferences naturally

  • Refining results through conversation

  • Presenting options in a clear, scannable format

The design challenge here was not just usability, it was clarity. Users needed to feel in control, even when interacting with an AI system.

Finally, trust played a critical role. When users interact with AI, uncertainty increases. To address this, we designed responses that were transparent, helpful, and consistent, avoiding overly “smart” behavior that could confuse or mislead

This project reinforced an important shift:
interfaces are no longer just screens, they are conversations.

Designing for this new paradigm means thinking less about flows and more about intent, context, and trust.

Designing AI & LLM Commerce with Google & Meta

Design & Tech

Apr 2025

other AI Delivery