Generative AI Tool
Case Study
Link to Figma file available upon request
Introduction:
Our main task was to create a generative AI tool with a dashboard to inform certain user groups of a multitude of usage from spending, to time spent and queries.
Internal tools to be used by everyone in the company, as well as moderators to keep informed on who was using which versions, as well as user management.
Problem Statement:
As a company, we have no way of understanding who is using our platform, how much it costs to run, what people are searching for, or how to improve their experience. Our goal is to provide the tools and information to groups to be able to track and predict usage to keep costs down while improving the experience and provide useful prompts to different user groups, while also creating a product that can be reused to sell for other companies.
Research:
Stakeholder Interviews
User Interviews
Whiteboarding sessions with expanded teams to gather requirements.
Using the data gathered from research methods above to inform design decisions with dot voting and smaller sessions with direct team to meet expedited timelines.
Design Process:
The first iteration of design was cleanup from the previous designer, implementing more reusable elements with proper frames and auto layout.
Creating a hybrid design system of previously implemented usage while creating new symbols and objects to reflect the needs of the product.
Daily design reviews and updates for different visual methods needed to comply with the current tech stack being used.
Solution:
Key features include user management solutions with Active Directory, deciding on data visualization tools, and updated custom satisfaction score implementation, and a tool to create custom themes.
Results:
Currently still in development on a more advanced MVP.
Reflection:
There have been more than 4 versions of the product that have been released so far with a cycling of designers, analysts, and developers while everyone is in between other client work.
This was an extremely quick product to iterate on. Multiple teammate switches, as well as quick feedback from shareholders while also expecting a product in a very short turnaround.
More focus and a singular solid team would have improved the outcomes, but you work with what you’ve got.