ADHD Scholarship Copilot

adhd-scholarship-copilot

Inspiration

Students with ADHD face a double-barrier when applying for scholarships:

Scholarships are written like legal documents full of dense text, unclear requirements, and overwhelming instructions.
Executive-function demands are high: switching tabs, tracking deadlines, understanding eligibility, gathering documents, and staying focused long enough to finish.

Because of this, hundreds of thousands of dollars in scholarships go unclaimed every year in Canada. Many qualified students never even apply.

We wanted to build something that reduces cognitive load, breaks tasks into steps, and helps ADHD learners stay engaged long enough to finish. Everything runs directly inside the browser.

What it does

The ADHD Scholarship Copilot is a Chrome extension that transforms chaotic scholarship pages into an ADHD-friendly workflow.

1. Scholarship Summary

Captures visible text from the page
Extracts deadlines, references, and key requirements
Generates a clean summary that reduces overwhelm

2. Eligibility Checker

Compares scholarship requirements with a saved student profile
Flags matched items: citizenship, program, degree type, and more
Flags unclear or missing information
Helps students decide quickly whether to apply

3. Autofill From Profile

Automatically fills common fields like name, address, school, program, graduation date, and citizenship
Reduces repetitive typing and mental fatigue

4. Focus Mode for ADHD

Provides a spotlight style focus shield
Supports 20 minute focus blocks with check-ins
Includes a calming break activity

5. Scholarship Library

Saves scholarship pages for later
Allows users to open, review, and delete saved scholarships
Keeps application tasks organized

How we built it

Frontend (Chrome Extension)

popup.html, popup.js, profile.html, profile.js
Chrome APIs for storage, scripting, and runtime messaging
Dynamic UI for summaries, eligibility checks, autofill, and the library
Focus shield implemented through injected scripts that track cursor movement and mask the screen

Backend (FastAPI)

A local FastAPI server on localhost:8000 with endpoints:

/parse for extracting scholarship requirements
/eligibility for comparing profile information
/bookmark and /bookmarks for saved pages
/plan reused from focus mode

LLM (Claude 3.5 Sonnet)

JSON safe prompt formatting
Extracts requirements and key values
Supports eligibility checking
Performs light reasoning to detect citizenship, undergraduate wording, first year completion, and similar patterns

Everything is factual, reproducible, and aligned with the actual features built.

Challenges

1. Autofill Detection

Different websites use inconsistent labels, placeholders, and structures. We built heuristics that inspect:

label text
placeholder text
nearby div text
input name and id attributes

2. Eligibility Reasoning

Requirements are written inconsistently across sites. We built reliable detection patterns for:

citizenship
membership categories
ambiguous terminology such as "penultimate year"

3. ADHD Friendly UX

Students with ADHD become overwhelmed easily. We had to balance simplicity with functionality and present only the most important information at each step.

4. Chrome Extension Complexity

We encountered challenges with:

dynamic script injection
messaging reliability
browser permissions
keeping the popup responsive without requiring reloads

Accomplishments we are proud of

Built a fully functional Chrome extension rather than static mockups
Implemented autofill, eligibility checking, and summarization
Built an ADHD-first focus system with spotlight and check-ins
Designed a calming interface that reduces cognitive load
Enabled students to understand a scholarship in under 20 seconds
Created a library system for tracking potential applications

What we learned

ADHD friendly design requires fewer steps, predictable interactions, and simplicity
LLMs can be powerful for summarizing but must be carefully constrained
Browser extension environments have many edge cases
Requirement parsing varies more than expected from page to page
Testing on real scholarship pages is essential because every page behaves differently