Retail · Consumer AI

AI Shopping Companion.

A chat that composes menus while you walk the aisles, translating cart into meals, budget into choices.

Sector · Retail · Consumer AI Surface · Mobile chat Status · Live consumer product

The problem

Grocery shopping is two problems pretending to be one: what do I cook this week, and what do I put in the cart now. Most apps solve one or the other. Meal-planning apps don't know what's in your cart. Shopping-list apps don't know what you can cook with it.

Our approach

A conversational companion that follows the user through the store. As items go in the cart, the assistant composes menus from what's already chosen and suggests what to add to complete them. The conversation is the interface, the user doesn't fill in forms, they think out loud.

The hard part isn't the LLM. It's the latency, the offline tolerance, and the model of "what makes a coherent menu", which has to feel native to how the user actually eats, not how a recipe site thinks they should.

What we shipped

  • Conversational shopping companion on mobile
  • Real-time menu composition from cart contents
  • Budget-aware suggestion logic
  • Live consumer product

Why it works

The product meets the user where they are, in the aisle, with a phone, with seven minutes of attention. Not at home with a recipe book and a plan.

← Back to all work

Conversational AI in a real-world workflow?

The LLM is the easy part. Latency, tolerance for being wrong, and matching the user's mental model is the hard part.

Book a call