Marvel AI

Marvel AI is an open-source AI Teaching Assistant designed to streamline educators' workflows, helping educators generate contextual content for lesson planning, grading, testing, tracking academic deadlines, and more.

https://app.marvelai.org/

Deliverables

Product Audit 
Market Research
Stakeholder Interviews
User Journey Mapping
Feature Wishlist
Information Architecture
Lofi/Hifi Prototype
Wireframing
Design Systems
User Testing
Project Hand-off

Goals

User flows supporting educator workflows, making sure it’s intuitive and accessible
Business user satisfaction through all-in-one solution for educational planning
Product customizing functionality for educators—elevating the overall experience 

Pain Points

Basic UI current UX is too simple to capture the nitty-gritty of educator workflows
Expectation Gap the GenAI software does not produce results as expected teachers
Oldschool educators prefer to have more control and  not so open to AI workflows

Research Insights

Need
While there are popular Gen AI, only a few other products solved for the specific use case of teachers

Conducted competitive analysis and user research through workshops and case studies to identify pain points through these tools

User Insights
Revealed user frustration with AI flows highlighting the need for more flexible, guided input mechanisms and context-specific customization

Need for real-time collaboration and refinement of prompt through integration of live editor, conversational assistance, and AI suggestions

1. Settings Page

2. Rubric Generator

Key Considerations
Format and templates
Accessing rubrics from previous years
Live editing of input






3. MCQ Generator

Key Considerations
Familiar text editing UI
AI smart suggestions
Global and local edits

3. Image Generator

Key Considerations
Feature cards and smart suggestions
Input breakdown for better prompting
Regeneration and edit prompt features

Colors, Type, Buttons and Icons
Built on existing design systems and a preference for dark mode. Established a design system with a defined primary and secondary colors and typefaces by creating color variables and text styles in Figma, enabling team members to build adjacent features with a cohesive brand image. Standardized buttons, grids, icons, and components were documented in a shared library for easy reuse.

User Success Measurement

Conducted user testing through A/B tests, surveys and interviews to iterate and developing refined version

Tweaking aspects like number of image generated, editing features, button placements, UX copy, info hierarchies

Defining success through data analytics, customer retention and clicks

A/B tests for toolbar with/without hover