Ask Rollie
Goal
Create an AI-powered chat experience that lets Onaroll operators quickly query their own performance data—without needing third-party analytics tools. The aim was to make complex data accessible in seconds and empower store leaders to make faster, data-driven decisions.
Role
Senior Product Designer
Tools
Figma, OpenAI (ChatGPT) API, internal Onaroll data platform
Overview
Context
Onaroll wanted to make store data more accessible for operators who rely on it every day. While the company already had powerful dashboards, most insights were hidden behind complex filters and reports. Ask Rollie was created to make that data conversational — letting users ask questions in plain language and get instant answers.
Challenge
Operators often knew what they needed but didn’t know where to find it. The goal was to design a lightweight, AI-powered chat that could translate everyday questions into real, usable insights. It had to feel natural, friendly, and intuitive while working within MVP constraints.
Role
As the Senior Product Designer, I led end-to-end design from research to delivery. I partnered with product, data, and engineering to define the experience, test flows, and create a simple interface that made complex data feel easy and human.
Design Process
Understanding Users
I started by talking with store operators and managers to understand how they searched for data and what made the process frustrating. Most people knew what they wanted to find but didn’t know where to look. That insight became the foundation for Ask Rollie — a tool designed to make getting answers as simple as asking a question.
Exploration and Iteration
I explored different ways to blend a familiar chat experience with the complexity of displaying data. Early concepts tested how messages, tables, and summaries could fit together without feeling overwhelming. Through several rounds of feedback, the design evolved into a clean, conversational interface that kept the focus on clarity and ease of use.
Collaboration and Validation
I partnered closely with product, engineering, and data to shape what was possible for the MVP. Together we defined how ChatGPT would generate responses, refined interaction details, and tested the flow with real users. The final design delivered an experience that made insights fast, friendly, and accessible for everyone.
Outcome
Impact
Ask Rollie made data feel accessible for everyone. Operators could finally get answers to everyday questions without digging through dashboards or waiting on reports. Early feedback described it as “like texting your data,” and teams quickly adopted it as part of their daily routine.
Results
The MVP successfully proved the value of conversational data tools, reducing the time it took to find key insights and improving engagement with performance metrics. It also opened the door for future AI-driven features across the TMX platform.
Reflection
This project taught me how powerful simplicity can be — especially when designing for users who are busy, mobile, and not always tech-savvy. It also reinforced the value of working closely with cross-functional partners to turn ambitious ideas into something tangible and genuinely useful.