A11Yson

We make the web easy!

Comment

Inspiration

The inspiration for A11Yson came from a simple observation: the web is technically accessible in many places, but still deeply uncomfortable for a large number of people.

Through research into WCAG guidelines and inclusive design principles, we noticed a recurring gap between compliance and experience. Many websites meet accessibility standards on paper, yet still overload users visually and cognitively through dense layouts, harsh contrast, motion, and cluttered interfaces.

We weren’t interested in “fixing” users or creating a charity-style accessibility tool. Instead, we wanted to explore how accessibility could act as a level-setter, removing unnecessary friction so everyone starts from the same baseline. That mindset led us to A11Yson.

What it does

A11Yson is a cognitive-first accessibility engine delivered as a Chrome extension.

It dynamically adapts how any website is presented based on a user’s needs, without removing content, rewriting meaning, or breaking the original layout. A11Yson focuses on presentation, not control.

Users can choose from four research-informed presets:

  • ADHD – reduces cognitive load and visual distractions
  • Dyslexia – optimizes typography and spacing for reading clarity
  • Sensory – softens contrast, reduces motion, and limits visual intensity
  • Clean – a universally accessible, minimal baseline

A11Yson also includes an AI conversational companion that can summarize page content, answer questions, or read information aloud, giving users multiple ways to access the same content.

How we built it

A11Yson was built as a three-layer accessibility ecosystem:

Intelligence Layer (FastAPI + Gemini 2.0 Flash) We use LLM-driven analysis to interpret user onboarding responses and generate structured accessibility profiles. Gemini also powers real-time summarization to reduce information density.

Command Center (Next.js 15) The dashboard acts as the source of truth for user profiles and settings, syncing changes across tabs and sessions using secure browser messaging and cloud storage.

Execution Engine (React Chrome Extension) The extension performs real-time DOM mutation using CSS injection and JavaScript logic. This allows us to adjust fonts, contrast, spacing, motion, and imagery instantly on any webpage.

Our system was designed to be modular, scalable, and respectful of existing web content.

Challenges we ran into

One of the biggest challenges was balancing technical power with restraint.

It’s easy to over-modify a page and accidentally remove meaning, context, or functionality. We had to be intentional about making changes that reduce friction without taking control away from the user.

We also navigated technical hurdles like Content Security Policy (CSP) restrictions, cross-tab state synchronization, and ensuring low-latency AI interactions that felt seamless rather than disruptive.

Design-wise, avoiding a “medical” or patronizing tone while still communicating accessibility clearly was an ongoing challenge.

Accomplishments that we're proud of

Building a real-time accessibility system that works across any website

Framing accessibility as equity, not accommodation

Successfully integrating AI in a way that supports WCAG’s “multiple means of access” principle

Creating visible, dramatic before-and-after transformations without altering content

Designing presets that are grounded in research rather than stereotypes

What we learned

We learned that accessibility is not a checklist, it’s a relationship between humans and systems.

Small visual decisions compound quickly, and cognitive load is just as important as visual clarity. We also learned that inclusive design becomes more powerful when users remain in control, with the ability to opt in, opt out, and customize freely.

From a technical standpoint, we learned how to architect systems that bridge static content with dynamic human needs, and how to responsibly integrate AI into that process.

What's next for A11Yson

Next, we want to expand A11Yson beyond presets into context-aware adaptation.

Future plans include:

  • Automatically detecting cognitively demanding pages and suggesting support

  • Image understanding and description for visual content

  • Smarter, adaptive profiles that evolve with user behavior

  • Additional presets for other access needs, including low vision and motor impairments

Our long-term goal is to make accessibility invisible until it’s needed, and powerful when it is.

Built With

Share this project:

Updates