Decipher.io
A Gift for Creative Thinkers
The problem Decipher.io solves
Many people with dyslexia and reading difficulties struggle to access written content not because they lack understanding, but because most digital material is not designed for neurodiverse readers. Dense PDFs, academic papers, and cluttered web pages cause letters to appear jumbled, increase cognitive overload, and make reading slow and exhausting.
Existing accessibility tools usually solve only one part of the problem - such as font changes or text-to-speech - while ignoring language complexity, visual stress, focus loss, and the reliability of AI-generated content. This forces learners to reread excessively, depend on external help, or avoid long-form content altogether.
Decipher.io solves this by transforming PDFs and web pages into a fully dyslexia-friendly reading experience through:
-
AI-powered content refinement and summarization : converts complex text into simpler language with adjustable difficulty levels, along with concise summaries for faster understanding.
-
Comprehensive visual customization: includes dyslexia-friendly fonts, font size, letter spacing, line height, color themes, and visual overlays to reduce eye strain and cognitive load.
-
Focus-enhancing reading: tools such as Bionic Reading, Focus Band, and sentence-wise Focus Mode to improve attention, reading flow, and comprehension.
-
Multimodal learning support: enabling text-to-speech with customizable voice and speed, along with real-time voice-based AI assistance for doubt solving.
-
Trust-aware AI design: using AI confidence indicators to highlight low-confidence refinements and reduce hallucination risks.
-
Accessibility across platforms: allowing users to download dyslexia-friendly PDFs and use a browser extension to convert any webpage into an accessible format.
By adapting content to the reader’s needs, Decipher.io enables independent learning, better comprehension, and inclusive access to knowledge across documents and the web.
Challenges we ran into
Building Decipher.io came with multiple technical and design challenges, especially due to the sensitive nature of accessibility-focused tools.
-
Extracting clean text from PDFs was tricky due to complex layouts, inconsistent formatting, and broken reading order, which directly affected AI refinement quality. We handled this by implementing layout-aware parsing and post-processing to clean, reorder, and normalize text before sending it to the AI pipeline.
-
Maintaining the right level of AI simplification was difficult, as over-simplification risked losing important meaning while under-simplification remained hard to read. This was addressed by introducing adjustable difficulty levels and refining prompts to preserve semantic accuracy.
-
Ensuring trust in AI-generated content was critical, especially for educational use cases, where hallucinations can mislead learners. To tackle this, we added AI confidence indicators that highlight low-confidence outputs instead of presenting AI results as absolute.
-
Designing for diverse dyslexia needs required careful handling of visual stress and cognitive load, since no single design works for all users. We solved this by building a modular, toggle-based interface that lets users customize fonts, spacing, overlays, and focus tools as needed.
-
Real-time voice interaction and webpage transformation introduced performance and consistency challenges. These were mitigated by optimizing asynchronous voice processing and using Document Object Model-based filtering in the browser extension to extract and transform readable content efficiently.
Overall, these challenges pushed us to focus not just on building features, but on creating a responsible, inclusive, and user-first solution, where accessibility, trust, and real-world usability were treated as core design principles rather than afterthoughts.
Tracks Applied (1)