Ongoing Project

Kumu Connect

Kumu: Teacher in ʻōlelo Hawaiʻi.
Empowering Hawaiian computer science education with culturally informed asset-based AI tools for teachers.

This project is part of the the educational efforts of Ka Moamoa Lab, in collaboration with a K-9 Hawaiian Immersion Language School in Oahu, Hawaii.


The partnership focuses on enriching students' learning experiences through local Hawaiian language and culture (place-based education).

TEAM

This project in collaboration with educators, researchers, designers, scientists, and community partners from within and outside the Native Hawaiian community, whose support is essential in this project.


  • Teachers and staff from partner school

  • Rachel (Lead UX Designer)

  • Will (UX Researcher)

  • Leah (UX Researcher)

  • Jahnavi (UX Desginer)

  • Dr. Josiah Hester and Dr. Judith Uchidiuno (Lab Advisors)

  • Collaborators from Hawaii Dept of Education (HIDOE)

TIMELINE

This project was conceptualized in August 2023 by and is ongoing with design and development efforts today.

ROLE

Competitor analysis, ideation and prototyping, contributed to design system.

TOOLS

Figma, FigJam, Qualtrics, Notion, Dovetail

TEAM

This project in collaboration with educators, researchers, designers, scientists, and community partners from within and outside the Native Hawaiian community, whose support is essential in this project.


  • Teachers and staff from partner school

  • Rachel (Lead UX Designer)

  • Will (UX Researcher)

  • Leah (UX Researcher)

  • Jahnavi (UX Desginer)

  • Dr. Josiah Hester and Dr. Judith Uchidiuno (Lab Advisors)

  • Collaborators from Hawaii Dept of Education (HIDOE)

TIMELINE

This project was conceptualized in August 2023 by and is ongoing with design and development efforts today.

ROLE

Competitor analysis, ideation and prototyping, contributed to design system.

Figma, FigJam, Qualtrics, Notion, Dovetail

TOOLS

CONTEXT

Reframing UX Practices

User Problem

Throughout this project, we critically examine standard UX methods and learn from Indigenous scholars about appropriate research practices that elevate Indigenous knowledge systems, promote equity, and foster co-creation with the community.

CONTEXT

Reframing UX Practices

Throughout this project, we critically examine standard UX methods and learn from Indigenous scholars about appropriate research practices that elevate Indigenous knowledge systems, promote equity, and foster co-creation with the community.

User Problem

Hawai'i Department of Education mandates Computer Science (CS) education next year, but most teachers don't feel confident teaching it.

Place based solutions rely on low tech solutions with ʻōlelo classes having even limited teaching resources.

Schools face a frequent need for substitute teachers, causing students' learning discontinuity.

CURRENT SOLUTION

The current solution is the 2nd iteration based on teachers feedback. We are currently working out the development details based on which the design will be revised.

Kumu Connect allows teachers to efficiently contact substitute teachers, through input fields tailored to their class and administrative needs.

Use Kumu Connect AI to generate plans that are place-based and culturally informed.

Edit sub plans using the platforms Document Editor that aligns with teachers' mental models when creating teaching materials.

IMPACT

  1. Accurately identified pain points in the user journey - helping teachers save time, feel less stressed, and more equipped when creating sub plans.

  1. We were able to gather our findings and give back to community based by writing an academic paper.

  1. We were able to gather our findings and give back to community based by writing an academic paper.

  1. Secured funding from the Google Academic Research Awards to build the platform, and the team is currently working on developing an LLM for ʻōlelo Hawaiʻi.

“With the sub plans, it would take me at least an hour [to create one] for one day's worth of plans. That took 10 seconds. And if I'm sick, this would be amazing.”

IMPACT

  1. Accurately identified pain points in the user journey - helping teachers save time, feel less stressed, and more equipped when creating sub plans.

“With the sub plans, it would take me at least an hour [to create one] for one day's worth of plans. That took 10 seconds. And if I'm sick, this would be amazing.”

  1. We were able to gather our findings and give back to community based by writing an academic paper.

  1. Secured funding from the Google Academic Research Awards to build the platform, and the team is currently working on developing an LLM for ʻōlelo Hawaiʻi.

HERE'S HOW WE GOT THERE

Design Process

RESEARCH

Goals

Methods

01 Establishing a knowledge foundation of history and culture

02 Understanding what Hawaiian Language Immersion experience looks like

03 What assets do teachers leverage in the classroom?

04 Understanding Computer Science Education

Secondary Research: Resources shared by partner school teachers, history documentaries, current events, and academic papers.

Talk Story with teachers and classroom observations.

Secondary Research: HI Department of Education: Statewide Survey on CS Education

Goals & Methods

Analysis

Affinity Mapping

Qualitative Coding

Assets Analysis

RESEARCH

Goal 1 Establishing a knowledge foundation of history and culture

Secondary Research: Resources shared by partner school teachers, history documentaries, current events, and academic papers.

Goal 2 Understanding what Hawaiian Language Immersion experience looks like

Goal 3 What assets do teachers leverage in the classroom?

For this, we Talked Story with teachers and classroom observations.

Goal 4 Understanding Computer Science Education

Secondary Research: HI Department of Education: Statewide Survey on CS Education

Analysis

Affinity Mapping

Qualitative Coding

Assets Analysis

DEFINE

Key Findings

observation, interviews

observation, interviews, lit review

conference, lit review

conference, observation

F1. Stand-in teachers have varied backgrounds

F1. Stand-in teachers have varied backgrounds

F2. Frequent subbing creates discontinuity in learning

F2. Frequent subbing creates discontinuity in learning

F4. Place-based schools lean on low tech solutions

F4. Place-based schools lean on low tech solutions

F5. Teachers are open to new tools that improve efficiency

F5. Teachers are open to new tools that improve efficiency

F3. Most teachers do not feel confident teaching CS

F3. Most teachers do not feel confident teaching CS

F6. The ʻŌlelo branch has fewer teaching resources

F6. The ʻŌlelo branch has fewer teaching resources

interviews, lit review

observation, interviews

DEFINE

Key Findings

observation, interviews

F1. Stand-in teachers have varied backgrounds

observation, interviews, lit review

F2. Frequent subbing creates discontinuity in learning

F3. Most teachers do not feel confident teaching CS

interviews, lit review

conference, lit review

F4. Place-based schools lean on low tech solutions

conference, observation

F5. Teachers are open to new tools that improve efficiency

F6. The ʻŌlelo branch has fewer teaching resources

observation, interviews

DESIGN

Concept Sketching

1. Smart Search

  1. Class Coverage

  1. Class Compass

We explored several AI-driven solutions for supporting substitute teachers. I ideated concept 1 feedback for which surfaced accessibility of laptops for and the time demands for lesson prep. Teachers preferred a centralized communication platform (concept 2) that facilitates sub-to-teacher arrangements with AI-assisted, editable lesson plans (concept 3).

User Flow & Wireframes

Concept testing led to a focus on streamlining the user flow for requesting a sub and preparing a sub plan when a teacher is unexpectedly absent. Feedback on this user flow confirmed the relevancy of this approach, integrating with existing sub binder practices and not requiring substitutes to have laptops.

Laying the Design System

I helped establish the design system, including a calendar dropdown that allows for selecting multiple dates and an AI prompt input modal for specifying the tone and purpose of content to be generated.

The typeface used also supports accents, which enables the platform to support multiple languages effectively.

Design Iterations

Using our design system, we created screens and user flows for three tasks: creating a sub request, prompting AI to generate a sub plan, and modifying activities in the sub plan. After in-person testing and analyzing both quantitative and qualitative feedback, we developed the second and current version of the design iteration.

Task 2: Prompt AI to Create Sub Plan

1st Iteration

Feedback insights:

(-) Teachers struggled to remember which class and subjects to prompt for when creating a binder.

(-) They needed more guidance for prompting the AI.

(+) They appreciated how the solution simplifies the sub creation process, allowing them to start rather than begin from scratch.

2nd Iteration

Introduced a new side panel that summarizes the class information for which a sub binder needs to be created.

We introduced example prompts to provide clearer guidance for users when creating AI prompts.

We introduced example prompts to provide clearer guidance for users when creating AI prompts.

Task 1: Creating a Sub Request

Feedback insights:

(-) Teachers were confused about what requests were active, pending or covered.

(+) Olelo to English and calendar view toggle

1st Iteration

2nd Iteration

Changing the layout to vertical sections allowed for more space and functionality to add filtering requests tabs.

Using colored tags in the cards to convey the status of the request.

Task 3: Change Activities in Sub Plan

Feedback insights:

(-) The previous sub plan editor mimicked a physical binder, causing confusing visual hierarchy and misinterpretation of left-side resources. AI editing was also buried, requiring clicks on individual activities.

(+) Teachers praised the generated sub plan for its step-by-step guidance and class-specific resources.

1st Iteration

The layout is now centered on the sub plan, with navigation in a side menu and AI assistance in a floating right-side module.

Manual editing now mirrors the familiar document editor model, enabling users to edit directly and intuitively.

2nd Iteration

We improved the print and share flows by following the zig zag eye grazing pattern, a high-priority feature because it ties into existing assets used by teachers. This addressed a usability issue where teachers struggled to locate the share/print button.

2nd Iteration

1st Iteration

DESIGN

Concept Sketching

We explored several AI-driven solutions for supporting substitute teachers. I ideated concept 1 feedback for which surfaced accessibility of laptops for and the time demands for lesson prep. Teachers preferred a centralized communication platform (concept 2) that facilitates sub-to-teacher arrangements with AI-assisted, editable lesson plans (concept 3).

1. Smart Search

  1. Class Coverage

  1. Class Compass

User Flow & Wireframes

Concept testing led to a focus on streamlining the user flow for requesting a sub and preparing a sub plan when a teacher is unexpectedly absent. Feedback on this user flow confirmed the relevancy of this approach, integrating with existing sub binder practices and not requiring substitutes to have laptops.

Laying the Design System

I helped establish the design system, including a calendar dropdown that allows for selecting multiple dates and an AI prompt input modal for specifying the tone and purpose of content to be generated.

The typeface used also supports accents, which enables the platform to support multiple languages effectively.

Design Iterations

Using our design system, we created screens and user flows for three tasks: creating a sub request, prompting AI to generate a sub plan, and modifying activities in the sub plan. After in-person testing and analyzing both quantitative and qualitative feedback, we developed the second and current version of the design iteration.

1: Creating a Sub Request

1st Iteration

2nd Iteration

Changing the layout to vertical sections allowed for more space and functionality to add filtering requests tabs.

2: Change Activities in Sub Plan

1st Iteration

2nd Iteration

The layout is now centered on the sub plan, with navigation in a side menu and AI assistance in a floating right-side module.

  1. Print Plan Flow

1st Iteration

2nd Iteration

We improved the print and share flows by following the zig zag eye grazing pattern, a high-priority feature because it ties into existing assets used by teachers. This addressed a usability issue where teachers struggled to locate the share/print button.

REFLECTION

1

Relationships and trust matter, both within the team and with our partner school. This foundation of trust and understanding was the bedrock of the project's impact.

2

There is a balance between breaking the grid in design and adhering to good design practices. Innovative design needs testing and refinement to ensure usability.

3

Identifying what is possible and available to build within the environment is as important as having a future vision. Understanding what can be integrated into existing systems is crucial, rather than introducing technology thar can be perceived as invasive.

1

Relationships and trust matter, both within the team and with our partner school. This foundation of trust and understanding was the bedrock of the project's impact.

2

There is a balance between breaking the grid in design and adhering to good design practices. Innovative design needs testing and refinement to ensure usability.

3

Identifying what is possible and available to build within the environment is as important as having a future vision. Understanding what can be integrated into existing systems is crucial, rather than introducing technology thar can be perceived as invasive.

REFLECTION

Designed with care, coffee and countless all-nighters.

JAHNAVI KOLAKALURI © 2024