XPRT.ai — UX case study
Business solution for the Advance data annotation mobile platform.
XPRT Tool is an advanced annotation tool, Xprt.ai that serves both, the labelers and project managers. It reduces their workload, maximizes accuracy, and subsequently improves the rate of algorithm success.
Annotate data quickly and earn rewards with the XPRT.ai Data Annotation Platform.
In order to extract the desired results from Machine learning/AI technologies, developers require high-quality data labeling. A single mismatch in labeling can lead to inaccurate and unreliable models.
XPRT.ai is a mobile application for both iOS and Android that lets you recognize the data(Retail, Healthcare, Sport, Text …etc) and find matchable data, and It uses accurate data that can be combined by the user selection process with our redefined algorithms.
I added gamified approach to data annotation to get an accurate dataset right from the start as you set to develop ML training models. Also, Precisely annotate words, images, music, and videos for effective and accurate training of algorithms. It gives error-free results.
I decided to UCD(user-centered design) process to make sure that my design decisions focusing on empathizing and define the stage of the design thinking framework.
To begin my research, I started to look at a goal and target audience, and analyzing UI, UX, User flow, IA, and key features. I dig into the detailed summary of the app's use and its features.
Also, I mainly focused on the app “To give simple and gamified AI Training model App for users and team managers in AI/ML industry”.
How might we deploying the world class-AI and deliver a high-quality training data model through our app?
- Provided a guided introduction to personalize the app
- Creating a gamified interface for user engagement
- Simply giving accurate labeling and data annotation application for users
While creating a user profile, we created an introduction to the app for data annotation users where they can glimpse the app look at once before they perform the labeling tasks.
users can annotate images, video, audio, text data on the mobile platform.
I have used the “Usability-hub” tool for the usability test and direct calls using “Zoom” with the internal team and stack holders. I have gathered the data from Usability-hub and identified the pain points of the users.
Questions asked in the interviews
- Can we schedule a task for every month’s first week or every week for 4 months?
- Can we update leaderboard points every day for multiple users?
- What type of features are you expecting in the rewards earning mobile application in Ai and ML industry?
- Would you like to get notified when a new task is available in the category?
Identifying Pain Points
I reviewed the data of the tests and noted down each user’s pain point onto a whiteboard. Then grouped the pain points into similar categories.
Then I prioritized each pain point based on its importance to the user as well as its importance to App. My assumptions of the importance to users were based on conversations with the users.
Defining the Problem
I decided to fix all the pain points that were important to Users and XPRT.ai. I redefined the pain points below.
1: Users had trouble finding how to start the task or how to start a healthcare category tasks
Users can start only from few categories after choose(some categories need degree certification or valid documents), which creates confusion for the users after choosing some categories like legal, medical.
This leads to the discoverability issue of the feature. People tend to look for their interesting category tasks from XPRT.ai mobile platform.
he only way they can do tasks by adding the valid certificate or documents of the category. i.e. if the user is having a doctor or a lawyer, he/she can do the task and earn rewards from that category after filling in the details from the profile screen. otherwise, the category is still waiting for approval from the app admin or administrator.
2: Users could not interested to do the task categories from the XPRT without gamification or participation leaderboards, It can be boring to do so.
Nowadays, Users will have interested to see their name in the apps. Users tend to attract the rewards or points table in every game environmental application. Here, We followed the gamification model to revisiting every user, and participated points can be seen by the user their interested category tasks done by themselves and see their points and super coins in the leaderboard whenever they log in. Also, they can redeem available points for e-gift cards on the redemption page.
3: Users can face difficulties (while they doing a special task or normal task) if they get any different questions that are coming from different data categories.
All types of users can do their task in the XPRT.ai in their free time. but if anybody can get random questions from random unlabel data which can coming from the storage data. sometimes which cannot be relevant to that category. that is why we created report button for users to report that question or comment it for better to know about their frustration when they get unrelative questions or answers in the tasks.
4: First-time users can view demo introduction to the mobile application in the Xprt.ai app
Prototyping and Validation
After 2 iterations, finally I created Hi-Fi mockups solution prototype and used Adobe XD to create a clickable prototype. I tested the prototype with 5 new individuals. Below are the Hi-Fi mockups of my final solutions of my design solutions.
Users had difficulty in understanding in few things. I have worked on following things again.
1: Reporting questions while they are doing a task
Users can do the task every day if they have any conflict to do any question in the task that can take negative user impact to the user. We are decided to add a report button in every question in the task UI Screens.
2: There is a difficulty for each section level to assign a badge for expert or regular user
While every user can improving their levels by doing daily tasks in the handpicked categories (like medical, legal, sports, etc).
The first time, I took sample level UI badges that are not up to the mark. The team and I had a discussion every day and after few iterations, it has finally done and approved by the stakeholders and review team.