What We Learned Building a Blind-Led Business

Generative AI @ the Intersection of Accessibility, Inclusion and Design

When I began redesigning the Gateway Navigation website in April, I thought it would take about a week.

Seven weeks later, I realized the project had become something entirely different.

Yes, we built a new website. But the process reshaped far more than the website itself.

We rethought our business model, clarified our services, learned modern WordPress development, explored multiple AI tools, and discovered new ways of combining technology with lived experience.

This post is the culmination of the posts we have published over the summer on the Gateway Navigation website. Those posts have explored where Gateway Navigation is heading, how our services are evolving, the growing role of accessibility in a rapidly changing technological environment, and our belief that lived experience must remain central to the work.

This post brings those ideas together by looking at the experience from a different perspective: what happened when I actually had to build the thing?

AI Is a Tool — Not a Replacement

One lesson stood out above all the others:

Artificial intelligence can be an extraordinary productivity and learning tool, but it is not a substitute for human judgment, critical thinking, lived experience, or the ability to recognize when a project is heading down a rabbit hole.

Throughout the redesign, I explored several AI platforms, including ChatGPT, Claude, Copilot, and others. I used them in different ways to help:

• organize and clarify ideas
• explore alternative approaches
• improve content structure and messaging
• identify potential accessibility considerations
• support SEO and content development
• explain unfamiliar WordPress concepts
• generate and troubleshoot code
• • accelerate the process of learning new technical skills

But every recommendation required evaluation

But there was an important qualification: AI-generated suggestions were starting points, not final answers.

Every recommendation needed to be evaluated against the actual requirements of the website. Code needed to be reviewed and tested. Accessibility claims needed to be considered in the context of real users and assistive technologies. Content needed to sound like Gateway Navigation rather than like a generic AI-generated company.
As a blind website developer and accessibility consultant, I came to see that AI’s greatest value was not making decisions for me.

As a blind website developer and accessibility consultant, I quickly realized that the most valuable role AI could play was not making decisions—it was accelerating learning.

As a blind website developer and accessibility consultant, I came to see that AI’s greatest value was not making decisions for me.

It was accelerating my ability to learn, experiment, compare possibilities, and solve problems.

Choosing Creation Over Automation

There are now AI-powered website builders that can produce an impressive-looking website from a relatively simple prompt.

Some of the results can be remarkably polished.

Others can feel glitzy or superficial.

But for this project, there was another issue: I didn’t want to be a spectator in the creation of our own website.

Generative AI is also not a perfectly consistent system. Its responses can vary depending on the model, context, prompt, available tools, and other factors. Models and products can also change over time.

That variability can be useful for exploration, but it can become frustrating when you need a process that is repeatable, testable, and accountable.

For my purposes, I wanted to understand why something worked, not simply accept that an AI system had produced it.

So I chose WordPress and the block editor.

The learning curve was steeper, but the trade-off was important: I retained control over the decisions that mattered.

• • Every heading
• • Every image
• • Every navigation choice
• • Every accessibility decision
• • Every piece of content

AI could help me investigate those decisions, but it did not make them for me.

AI became a knowledgeable assistant sitting beside me — not the designer sitting in my chair.

Accessibility Cannot Be Prompted Into Existence

One of the most important lessons from the project was that accessibility is not something that can simply be added by giving an AI the right prompt.

AI can identify potential accessibility issues. It can explain standards. It can suggest semantic HTML, keyboard interactions, alternative text, heading structures, and other techniques.

But a technically plausible answer is not necessarily an accessible experience.

Accessibility exists in the interaction between people, technology, content, and context.

That distinction matters.

As someone who is blind, I bring lived experience to the development process that an AI system does not possess. I know what it is like to encounter a navigation structure that makes sense visually but becomes confusing with a screen reader. I know the difference between something being technically present and something being genuinely usable.

That does not mean my experience represents every blind or low-vision user. It means that lived experience provides an essential perspective that needs to be combined with testing, standards, research, and feedback from other people.

Human Insight Matters

The redesign also benefited from thoughtful feedback from colleagues and collaborators.

Their perspectives helped challenge assumptions, identify areas that needed clarification, and ensure that the final website reflected more than technology.

It needed to reflect who Gateway Navigation is becoming.

That is particularly important because this redesign was never really just about changing colours, layouts, or WordPress blocks.

It was about making the website better communicate our evolution from our earlier work into a broader accessibility consulting and social enterprise model.

The website needed to connect our experience in accessibility, inclusive navigation, assistive technology, emerging technology, and training with the practical needs of the organizations we want to work with.

That required technology.

But it also required conversation, reflection, experience, and judgment.

What Seven Weeks Taught Me

The original one-week estimate was obviously optimistic.

But the additional time wasn’t wasted.

The seven weeks gave me an opportunity to learn modern WordPress development, experiment with AI-assisted workflows, reconsider how we describe our services, and think much more carefully about what it means to build a business around lived experience.

Perhaps the biggest lesson was that speed is not the same thing as progress.

AI can make some tasks dramatically faster. It can help you get unstuck and reduce the time required to learn unfamiliar concepts.

But faster is only better when you are moving in the right direction.

Sometimes the most valuable thing a person can do is stop, question an assumption, and ask whether the current approach still makes sense.

That is where human judgment matters.

A Better Question for Organizations

For organizations considering artificial intelligence, I don’t think the most useful question is simply:

Should we use AI?

A better question is:

How can AI help people contribute more effectively while keeping human experience, expertise, accountability, and judgment at the centre?

That is the approach Gateway Navigation intends to continue championing as we help organizations navigate the rapidly changing worlds of accessibility and emerging technology.

• AI will continue to evolve.
• The tools will change.
• The interfaces will change.
• The capabilities will change.

But the need for people to understand what they are building, who they are building it for, and whether it actually works will remain.

For Gateway Navigation, that means keeping lived experience at the centre while using emerging technology to extend what people can do — not to replace the people themselves.

Continuing the Conversation

This website redesign started as a technical project.

It became a business transformation.

And, in many ways, it became a lesson in how I want Gateway Navigation to approach technology in the future.

• Use the tools.
• Learn from them.
• Question them.
• Test what they produce.
• And know when to step back and trust human judgment.

How is your organization balancing human expertise with artificial intelligence?

We’d love to continue the conversation.