Objective
As a founding designer in The Technical Magic Company, I was in charge of leading the company's product development and end-to-end design. I was hired during the pre-seed phase as the 7th employee. The company's mission was to create a dynamic personal AI companion using the various AI large language models and generative AI models. These experimental releases would help us understand the rapidly evolving nature of the AI space in 2023.
Results
During my short time there, I was able to be a part of two product launches:
+ Launched a generative AI browser extension, web-based portal, and prompt creator tool from v0 to v1
+ Launched a PWA mobile app and a responsive web application from v0 to v1
If you're curious to know more about this product, visit: https://technicalmagic.ai/
Wispy 2.0
Below is a snapshot of the final product we shipped: a web-based consumer generative AI application. I designed a simple web-app that allowed us to generate text, images, and video using multiple GenAI models. We wanted a simple and easy solution to teach users how to "prompt engineer" using graphical user interface inputs.
Simplified List of Prompts (Powers)
Our original product had a large library of powers that were mostly unused. Our goal with Wispy 2.0 was to simplify them in order to experiment with full functionality
Graphical User Interface Input
With the research from our users, we learned chat is too much of an open-ended interaction model for users. Our goal for this product was to help people understand what prompt engineering is and aid in their inputs.
Prompt Customization & Model Selection
My goal was to simplify image customization options using the technical tools that we had: changing image generator models, using negative prompts, styles as guides.
Runtime & Model Customizer
We wanted to make sure that users didn't have to choose a model every time they invoked a power. This meant that an overall runtime for the app using specific models would be accessible in the Settings.
Wispy 1.0
This is the first product we launched as a company: a web-based catalog of AI prompts that connected to the web extension. The web extension used prompts that could crawl the content of the page in order to extract desired results. Our most popular "powers" used were: summarize page, recipe extractor, and summarize in a different language.
Gallery of Powers & Creator Portal
Our goal was to democratize creating AI prompts we called "powers". I designed an area of the web extension that allowed people to edit the existing powers we already had in order to customize it. The original gallery that we had was not scalable for public release so I did a full redesign of the gallery of powers that included user-generated content and a creator portal.
First Run Experience
I began this project by doing a heuristic evaluation in conjunction with a simple usability test with 10 panelists. After identifying the key areas of usability issues, we learned that the main concern was going from download, to opening the extension on the browser. The solution to this came in 2 steps: youtube tutorial (easy but terrible experience) then interactive tutorial (more difficult, but much better experience).
