zepto

Designing a Conversational AI Agent for Zepto: From Friction to Flow in Grocery Ordering

App name / Client

Zepto

My Role

UI/ UX designer

Industry

E-commerce

Platform

Figma

project image

Reimagining Grocery Shopping with AI: Zepto’s Conversational Agent

A seamless, human-first AI shopping assistant designed to help users reorder essentials, discover new products, and finish their grocery runs with ease.

Project Overview

  • Timeline: 2 weeks
    • Tools used: Figma, Canva, Fig jam, Notion
      • Goals: Improve task flow for users shopping under time pressure.

        Problem Statement

        Grocery shopping felt repetitive. Even loyal users had to scroll, search, and remember every time. Zepto was fast, but not helpful. It needed a smarter way to assist, one that felt more human and less like starting from scratch.

        Research

        Insights Panel 

        • Users want AI that “remembers”
          • ✅Friction comes from repetitive actions
            • ✅ Shopping is goal-based (e.g., "Dinner prep", "Weekly restock")
              • ✅ Users prefer proactive help over too many choices

                Design Process

                User Journey

                StageGoalCurrent Journey (Before AI)Pain PointsImproved Journey (With AI Assistant)Delight Points
                1. EntryOpen app to reorder groceriesOpens app → Navigates categoriesTime-consuming, repetitiveOpens app → Sees “Reorder last week’s list?” promptFeels understood instantly
                2. BrowseFind additional itemsBrowses categories manuallyGets distracted or forgets itemsAI suggests “Add usual fruits?” or “Low on milk?”Saves time, fewer forgotten items
                3. DecideBuild cart & finalize orderManually adds each itemCognitive load, decision fatigueAI builds cart with options to tweakLight, easy cart-building
                4. CheckoutFinish order quicklyReviews cart → CheckoutFeels unsure, checks againOne-tap confirmation with smart summaryConfident & quick
                5. Post-OrderFeel good about experienceWaits → Forgets what she addedLow recall, little engagementGets helpful summary & “Want reminders?” optionTrust builds, repeat assured

                Once I tested the first version, I noticed small things that needed work, users missed the AI prompt or didn’t know where to click.

                So, I went back and refined:

                • Made the AI prompt more visible, right at the top
                  • Used warmer, more conversational language
                    • Cleaned up the flow to remove unnecessary steps

                      Final UI Screens

                      Learnings 

                      • Designing for conversations requires empathy, every word matters.
                        • Predictive UX isn’t about showing more, but showing what matters.
                          • Users love when the app feels like it knows them.

                            Expected Impact

                            • Increase Order Speed: Reduce the time taken to place an order by 30-40%.
                              • Boost Daily Conversions: Improve conversion rates by approximately 25%
                                • Enhance User Retention: Increase retention through personalized experiences.
                                  • Reduce Cognitive Load: Simplify the ordering process, making it more intuitive.