- Identified the user story and, together with the RO, defined the MVP requirements.
- As part of the concept, I conducted a series of moderated studies to test interface hypotheses.
Solution:
- UI
About
- Based on the analysis of the operating patterns of competitors' AI services, I have developed several concepts for solutions for working with text data
- Reused the editor's UX/UI solutions
- UX
Research:
- Conducted a user survey to identify key use cases for AI text processing.
- Analyzed competitive solutions to assess best practices and shortcomings.
Goal:
- For users: help them quickly complete routine text-based tasks
- For businesses: close the competitive gap and increase product value
Problem:
- Lack of initial business and technical requirements
Final design:
- Defended the concept to stakeholders.
- After analyzing the technical requirements, I broke the task into phases.
- Drew the final mockups and described the development documentation.
CJM
The CJM lays out a basic scenario for working with an AI assistant. Job Stories and User Stories are specified.
User flow
The flowchart shows a scenario for sending a request to an AI assistant and inserting the generated result into a document.
Survey
Goal:
Identify the main scenarios for working with a document
Method:
Users are asked to complete a survey. The questions cover their field of work, job title, document workflows, experience with AI, and their attitude and level of trust toward AI.
Result:
The study surveyed a total of 350 respondents. The survey provided insight into the scenarios for interacting with AI in editors. The survey served as the basis for creating prompts for the feature's "prompt library" and "prompt designer."
Scenarios for working with a spreadsheet document
Scenarios for working with a text document
Research
As part of the AI implementation, research was conducted, much of it within the Pathway service. Some of this research is listed below.
Competitor analysis
Goal:
Define the concept of AI in the interface
Method:
I analyzed competitors' AI-powered products and identified three concept types:
- popover
- sidebar
- chat
Result:
The right dock panel concept was chosen because:
- the AI assistant should not overlap the document's content area
- the user should have access to conversations and chat history
- the AI assistant is used as a collaborator, working with the user on the document
Determining the entry point
Goal:
Check the readable entry point of the feature
Method:
First-click method. The user had to find the entry point to the AI assistant.
Result:
A button with an icon (already a pattern in AI products) and an accent color was immediately recognized by users and left no doubt when choosing.
Final design
Goal:
Determine the final design
Method:
The survey asked users to evaluate their preferences in terms of dialog box style and to select the final option from three.
Questions to respondents:
- How serious does this screen look?
- How modern does this screen look?
- Which option do you like the most?
- Which of the three options looks like you'd trust it to handle your document?
Result:
- Completed the full cycle of implementing an AI assistant into the product, from requirements gathering to release.
- We became the first team to release an AI assistant within the company.
- Shared our expertise with other designers from the AI working group.
