Solo Product Designer
8 Months - Apr to Dec 2020
Reduced Labor Costs by 4%
Process Stability Improvement by 5%
To comply with my non-disclosure agreement, I have omitted and obfuscated confidential information in this work. All information in this case study is my own and does not necessarily reflect the views of platform.
Equipment engineers develop AI projects using their domain knowledge in the manufacturing industry on no-code AI platforms to monitor plant operations, prevent unexpected shutdowns, and optimize human resources.
Engineers routinely inspect 500 machines; but false alarms often occur and add extra work.
The whole factory requires 500 AI applications (one for each machine), far beyond the only one AI engineer in the factory could cover.
This version is full of data science terminology since it was built by data analysts. Its usability needs to be improved.
I focused on redesigning the interface. From needs interview, I learned the purpose of the AI platform and defined users’ main pain point. I also set the design goal for potential users from a business aspect.
How to support equipment engineers in creating AI projects that can efficiently solve the lack of AI experts and prevent false alarms?
The Repair Report of Machines mobile version offers a user-friendly and efficient way to manage and track machine repairs, ensuring proper documentation, transparency, and streamlined maintenance processes.
Allow users to quickly identify variable information by adding a variable description field. While users are familiar with the main variables, my new design reduces their cognitive cost during the model-building process by providing related information.
In previous design, users viewed the data and chart in a small area horizontally. Since it is easier to discover the changes in data through vertical viewing, I expanded the explore data area vertically in the new design so that users can effortlessly observe data trends and distributions.
In addition, users can view and process data using the cross-filtering function, easily untangling complex data and outputting data reports.
Shorten the project time by integrating users’ technical languages on the interface.
Users can use autoflow to build several algorithm models with their domain knowledge.
I conducted several iterations to identify existing issues and improve design with practical solutions.
What is the main purpose of the platform? How to choose a model? What’s users’ motivations to use the platform? How can users successfully build AI projects?
Users of the previous version easily got lost in its navigation system. My test results show that users were often unsure about what to do next. I combined data science and user scenarios to improve the navigation bar.
To better guide users in utilizing the platform, so the main task was to help them know better about the modeling process. I redefined the main interaction design layout to accelerate the AI model-building process.
I designed the design system and the brand identity for the new platform, including transformation, attainability, goals, and other information.
Redesigned interface with the new design system.
I planned the testing process and set quantifiable goals for the product rollout. Subsequently, I analyzed product feedback from potential users and provided reference for project managers to prioritize the following development project.
On the first day of official product trial, the redesigned AI platform received positive feedback from many potential customers, who felt much easier to build AI projects on the new platform.
The platform saved 4% labor cost in the AI project building process and also improved manufacturing process stability by 5%
Asia Manufacturing Facility Manager
The platform helped people without an AI background like me quickly understand the process of building AI models.
Managers
We used to write programs to draw charts, but now the platform automatically generates all charts and greatly reduces report-writing time
Data Scientist
The biggest challenge for me in the project was to understand the AI field. I kept consulting the team and users about how they had used AI and their domain knowledge. I also introduced users’ perspectives and UI design to my team, which was composed by 90% data scientist. I’m honored to be a part of the redesign project, to learn from it, and to integrate AI into the manufacturing sector.
"We have established a low-threshold AI platform that allows equipment engineers with non-AI backgrounds to improve their work using AI technology."
