Inspiration

Imagine understanding concepts clearly, but every word on the screen feels jumbled, letters seem to dance, and finishing a simple paragraph takes twice the effort. This is the everyday reality for nearly 20% of the population living with dyslexia.

Many students with reading difficulties drop out of school not due to lack of intelligence, but due to the absence of inclusive learning systems. Most digital content - PDFs, academic papers, and web pages - is not designed for neurodiverse readers, leading to cognitive overload, visual stress, and slow comprehension.

We built Decipher.io to bridge this gap and make knowledge truly accessible. Inspired by the idea of “deciphering” complex text into clear, readable content, our goal is to empower dyslexic learners and anyone struggling with dense information to learn independently and confidently.

What it does

Decipher.io transforms PDFs and web pages into a fully dyslexia-friendly reading experience.

Key capabilities:

  • AI-powered content refinement & summarization : Converts complex text into simpler language with adjustable difficulty levels and concise summaries.
  • Comprehensive visual customization : Dyslexia-friendly fonts, spacing controls, color themes, and overlays reduce eye strain and cognitive load.
  • Focus-enhancing reading tools : Bionic Reading, Focus Band, and sentence-wise focus mode improve attention and reading flow.
  • Multimodal learning support : Text-to-speech with customizable voice and speed, plus real-time voice-based AI assistance.
  • Trust-aware AI design : AI confidence indicators highlight low-confidence outputs to reduce hallucination risks.
  • Cross-platform accessibility : Browser extension converts any webpage, and users can download dyslexia-friendly PDFs.

By adapting content to the reader - not forcing the reader to adapt - Decipher.io enables independent learning and inclusive access to knowledge.

How we built it

We extract text from PDFs using PDF parsing and OCR, and capture webpage content through a browser extension with DOM filtering. The extracted text is processed by AI models that simplify language, generate summaries, and adjust difficulty levels while preserving meaning. Finally, the refined content is rendered with dyslexia-friendly fonts, customizable visuals, focus tools, and text-to-speech support to enhance readability and comprehension.

  • Frontend : React, Next.js, Tailwind CSS, shadcn/ui
  • Backend : PDF parsing, OCR, Browser extension (DOM filtering)
  • AI Functionality : Gemini 3 Flash Preview, Google Cloud TTS API, OpenAI fallback API

Challenges we ran into

Building an accessibility-first AI tool presented unique challenges:

  • PDF text extraction issues : Complex layouts and broken reading order affected AI output quality. Solved with layout-aware parsing and text normalization.
  • Balancing AI simplification : Over-simplification removed meaning; under-simplification remained difficult. Introduced adjustable difficulty levels and prompt tuning.
  • Trust in AI outputs : Hallucinations can mislead learners. Added AI confidence indicators for transparency.
  • Designing for diverse dyslexia needs : No single layout works for all users. Built a modular, toggle-based customization system.
  • Performance challenges : Real-time voice interaction and webpage transformation required optimization. Used asynchronous processing and DOM-based filtering.

These challenges shaped our commitment to responsible AI and inclusive design.

Accomplishments that we're proud of

  • Built an end-to-end accessibility platform for PDFs and web content.
  • Designed a trust-aware AI system with confidence indicators.
  • Enabled independent learning for dyslexic and neurodiverse users.
  • Delivered real-world impact aligned with Sustainable Development Goal 4: Quality Education.

What we learned

  • Neurodiverse users require customizable, not one-size-fits-all, solutions.
  • Accessibility is not a feature - it must be a core design principle.
  • AI must be transparent and trustworthy, especially in education.
  • Performance and usability are critical for assistive technologies.
  • Inclusive design benefits everyone, not just the target group.

What's next for Decipher.io

  • Counselor Support Portal: Bridges the gap with one-click booking for verified therapists.
  • Mobile OCR App: Scans physical text (books, menus) to read aloud or reformat.
  • Gamified Progress: Boosts confidence with reading streaks and WPM tracking.
  • Support for multiple languages and regional content

Built With

Share this project:

Updates