Employee AI Assistant
Competitive, heuristic and survey analysis
Competitive analysis
I reviewed other generative AI programs and identified commonalities, strengths and weaknesses.
Most applications supplement the text input field with pre-set actions the AI can take.
Starting an interaction with Jasper
Starting an interaction with ChatGPT
Editing parts of the AI's output was a common capability, while the granularity of control varied.
Editing the output in Jasper
Some features were not found in all applications.
HuggingChat's toggle to search the web
Google Gemini presents alternative versions of output.
Heuristic evaluation of the current state
I reviewed the current state of the AI application and graded areas of improvement based on severity of impact to the user—from task-impeding to nice-to-have.
Examples of issues that prevent the user from completing their task.
Content is not organized or easy to digest.
Settings aren't remembered by the application.
Examples of issues that add significant effort to complete a task.
Edits can't be made to the output. A user would have to start over and re-generate the summary.
The technical descriptions are not helpful to most users.
Minor issues—like misleading or inconvenient UI—lengthen task duration.
Because of the copy button's placement, it's unclear what content will be captured. The sources must be copied manually.
There were opportunities to add new functions that increase value for the user.
Adding the capability of comparing different outputs on one screen would save users effort.
Survey
I drafted a survey to measure employees' use and perceived value of the AI application.
Questions collected info on which features of the application supported employees' workflows.
Problem statements
From the research I synthesized problem statements to guide design decisions and priorities. Problems can be grouped broadly into two categories:
Surface-level problems that are small individually, and can be addressed with low effort from the development team.
Functional problems that require restructuring of the application and more effort from the development team.
