Losing My Vision and My Ability to Code

I've lost a lot of my vision in the past two years. This isn't a surprise—I've been expecting and fearing this moment since I was four years old. When you grow up with a degenerative eye disease, you learn to cherish the sight you have for as long as you have it. But expecting something doesn't make it any easier when it finally happens.

For a long time, coding was a source of joy for me. But as my vision declined, my ability to engage in text-heavy work like coding and writing became significantly harder. Programming involves a lot of precision, and while small errors like a missing semicolon might be a minor inconvenience for a sighted developer, for me, it became a showstopper. I couldn't skim through code as easily, I couldn't quickly spot where things went wrong, and I found myself increasingly frustrated and exhausted trying to maintain my workflow.

Try out this simulator to experience a simplified version of central vision loss. The blur follows your cursor to simulate how macular degeneration affects the central field of vision while peripheral vision remains intact.

The reality of being a low-vision coder is that everything takes longer. Consider a simple everyday task like making a pot of tea:

• Grab the pot from the cupboard. If it's not where I left it, I have to ask for help because I'm not going to find it on my own.
• Dig through my teas, using a magnifying glass or my phone's camera to read the labels.
• Grab my water level detector from its drawer and place it over the side of the teapot.
• Pour boiling water and wait for the device to beep, indicating the water has reached the right level.
• Stop pouring and continue brewing my tea.

It's a small process, but when you multiply this tedium by every single task in a day, it adds up. The same principle applies to coding. Every misplaced character or indentation meant I had to slow down, recheck everything, and spend far more time debugging than actually developing.

It doesn't help that accessibility is often an afterthought in IDEs.

Enter Claude: AI That Actually Helps

Then I discovered Claude 3.7.

I had tried AI-assisted coding tools before, but most of them didn't significantly improve my workflow. Either they made errors that were just as hard for me to catch as my own, or they didn't provide meaningful improvements beyond basic autocompletion. But once I configured Claude properly, I was honestly shocked by how well it worked. It wasn't just a tool that completed lines of code—it was an actual assistant that handled the details so I could focus on the big picture.

Claude allows me to describe the structure of what I want, and it takes care of implementation details. It can:

• Generate full blocks of code based on my descriptions.
• Catch and fix syntax errors before I even need to worry about them.
• Help me refactor code without needing to strain my eyes.
• Explain what a snippet of code does in plain language, reducing my need to visually parse long blocks of text.
• Create accessible features faster and better than most devs could consider

And the best part? I can do this all without having to stare at endless lines of code, saving me eyestrain and future-proofing my ability to make cool programs. Win-win!

The Magic of Cursor's AI-First Workflow

This became even more powerful when I paired Claude with Cursor. Cursor's AI-first workflow transformed my experience because it meant I wasn't just coding with AI support—I was working in an environment designed for AI-driven development.

With Cursor:

• I could rely on AI to handle repetitive and vision-intensive tasks like formatting and syntax corrections.
• The workflow was structured around collaboration with AI, rather than requiring me to manually correct its output.
• Debugging became significantly easier since I could ask Claude to analyze my code and suggest fixes in a conversational way, rather than me painstakingly scanning for errors.

The result? Coding became accessible to me again. I could focus on logic, structure, and creative problem-solving rather than wrestling with typos, indentation issues, or formatting inconsistencies. I could move fast without feeling like my disability was holding me back.

Accessibility Isn't Just About Screen Readers

One of the biggest misconceptions about accessibility in tech is that it's all about screen readers. But accessibility is much broader than that. For people like me, who are in the low-vision category (which represents 70% of the blind community), screen readers aren't always the best solution. Instead, we need tools that allow us to interact with text-heavy environments in ways that minimize our visual strain.

Claude and Cursor provided that for me. They made it possible for me to code efficiently without relying on magnification tools or constantly zooming in and out of my editor. They allowed me to keep working in a field I love without the crushing exhaustion of trying to visually parse every little detail.

AI Isn't Just a Convenience—It's an Equalizer

For many sighted developers, AI-assisted coding is a nice-to-have feature. It speeds things up, helps with productivity, and reduces some of the more tedious parts of coding. But for me, AI isn't just a convenience—it's an equalizer.

Without AI assistance, my productivity had plummeted. I was constantly fighting against my limitations, and even though I knew I was capable, my disability meant that I couldn't work at the speed I wanted. AI changed that. With Claude and Cursor, I can develop applications, build projects, and contribute at a level I thought was slipping away from me.

The Future of AI and Accessibility

My experience with AI in coding is just the beginning. There are so many ways AI can revolutionize accessibility, and I hope to see more tools that:

• Adapt to the specific needs of low-vision users.
• Provide natural language explanations for code and technical concepts.
• Integrate seamlessly with accessibility tools rather than being built as an afterthought.

Tech has a long way to go in making software development fully accessible to blind and low-vision individuals. But for the first time in years, I feel like I have a seat at the table again. AI has given me the ability to code without barriers, and that's something truly revolutionary.

If you're a developer with a disability, or if you know someone who struggles with traditional coding environments, I highly encourage you to explore AI-assisted workflows. They might just change everything for you, like they did for me.