Parse

Inspiration

Standard PDFs create a barrier for neurodivergent individuals and people with disabilities. Current digital reading assumes a single format works for everyone. This ignores the diverse ways our brains process information. Our inspiration came from the Assistive Innovation Challenge prompt to build software that supports people with disabilities through accessibility, autonomy, and independence. We wanted to address the call to make existing digital experiences more accessible. We did this by taking the complex PDF format and making it adapt to the visual and cognitive needs of the user.

What it does

Parse is an accessible web reader that adapts complex documents to fit the specific cognitive and visual needs of the user. Users upload a standard PDF of up to 20MB. Our engine then reworks the content to ensure it supports the reader.

Parse offers personalized cognitive profiles to restore reading autonomy:

• Standard: A clean accessible baseline.
• ADHD: A focus mode that generates short summaries.
• Dyslexia: Simplifies text to a 6th-grade reading level paired with readable fonts and warm backgrounds.
• Low Vision: Provides highly condensed text for massive scaling and high contrast visuals.
• Sensory: Features a calm neutral tone and automatically removes intense language and bright colors.

By tailoring the reading experience, Parse improves the daily life of users and directly aligns with the hackathon Impact criteria.

How we built it

We built the front end and core application on Next.js to ensure a fast and accessible web experience. For document processing, we integrated the Google Gemini API. Instead of just extracting text, we used Gemini to restructure the document formatting based on the selected user profile. This approach allowed us to build a technical implementation that scales to the problem being solved.

Challenges we ran into

Extracting text from standard PDFs into clean data before sending it to the processing engine required careful handling of edge cases like unusual tables or columns. We also spent a lot of time refining prompts with Gemini. We had to get the system to rewrite complex concepts to a 6th-grade reading level for the Dyslexia profile without losing the original meaning. Creating a cohesive design system was another challenge. We needed to accommodate extreme high contrast for low vision while maintaining a calming palette for the sensory profile.

Accomplishments that we're proud of

We are proud to have built a functional prototype that demonstrates respectful and inclusive design. We moved beyond basic accessibility overlays to create a system that restructures the document format completely. By establishing four distinct cognitive profiles, we approached a common accessibility challenge in a unique and creative way.

What we learned

We learned a lot about how different neurodivergent minds process text and visual information. This project reinforced the Assistive Innovation Challenge core message. It taught us to think about how technology can be designed for inclusion from the beginning. We also deepened our understanding of the potential of large language models in the assistive technology space beyond standard chatbots.

What's next for Parse

We plan to expand the platform by adding multi-language support to help neurodivergent individuals globally. We also intend to develop a browser extension so that the cognitive adaptation engine can automatically rework any webpage or digital text. This will bring us closer to a world where technology is universally inclusive.

Built With

• groq
• next.js