For 2 years, I have created Blueprints using a screen reader. Like many blind users, I have become highly skilled at navigating complex applications, moving through pages, locating controls, and finding the information I need. It is simply part of how I work.
When I first heard about creating a Blueprint through AI chat, I was skeptical.
My initial reaction was that there was no way this approach would be easier. I assumed I would spend more time trying to understand what the AI was doing than actually building the Blueprint. I expected frustration. I expected confusion. Honestly, I assumed I would prefer the traditional approach.
After building an entire Blueprint with AI, I changed my mind.
More Than Just a Faster Way to Build
One of the biggest differences between the traditional Blueprint experience and the AI-assisted experience was that the conversation became part of my learning process.
In the traditional interface, I can complete tasks because I know where things are and how they work. However, the interface does not always explain why something should be configured a certain way.
With AI, I could ask questions as I worked.
If I was adding a step to a stage, I could ask why that step might be useful. If I was unsure about a particular configuration choice, I could ask for an explanation. The assistant didn't just perform actions. It helped me understand the reasoning behind them.
I found myself wishing I had access to this type of guidance when I first started at Pega. I believe I would have learned the Blueprint process much faster because I would have understood the purpose behind the configuration decisions rather than simply following steps.
The Learning Curve Was Real
My experience was not immediately positive, initially, I was actually less efficient using AI.
I had to learn how to interact with it. I needed to understand the pace of the conversation, how to ask questions effectively, and what kind of information the assistant needed from me. There were moments early on when I wanted to stop using AI and go back to creating the Blueprint manually because it felt faster.
But something interesting happened as I gained experience.
The more I used the assistant, the more natural the interaction became. I learned how to shape my requests, and the assistant became easier to predict. By the end of the testing, I felt that I was working more efficiently than I would have been manually.
More importantly, I wasn't just completing tasks. I learned throughout the process.
Why AI Was Different for Me as a Screen Reader User
The greatest benefit was not speed. It was navigation.
Traditional applications often require screen reader users to spend significant time moving around an interface, locating controls, finding the correct section, and keeping track of where they are on the page. Over years of using assistive technology, this becomes second nature, but it still requires effort.
With AI, much of that effort disappeared.
Instead of spending time searching for controls, I could remain focused on the conversation and ask the assistant to perform actions on my behalf.
This fundamentally changed how I interacted with Blueprint. Rather than navigating to accomplish work, I could simply describe my intent.
That may seem like a small difference, but for a screen reader user it can significantly reduce the cognitive effort involved in navigating a complex application.
Another unexpected advantage was spelling. I have never been a particularly strong speller. When creating case types or entering names, the assistant frequently understood what I intended and corrected minor mistakes automatically. It may sound like a small feature, but it removed another source of friction from the process.
Where the AI Assistant Fell Short
The experience was not perfect. One of my biggest frustrations was the inability to duplicate workflows through AI. Since workflow duplication can already be performed manually within Blueprint, I expected the assistant to support the same capability. Instead, I found myself repeatedly trying to convince the AI to perform an action it could not do.
I also encountered challenges when working with uploaded files. In one case, the only file available for testing was related to cars. When I attempted to add it to an existing case type, the AI refused because the file content did not match the case type. To complete my testing, I ultimately had to create a car-related case type simply so the assistant would accept the file.
There were also times when the AI needed information that I did not realize was required. For example, when working in Workflow Details, I eventually learned that I needed to specify which case type I wanted to modify before asking certain questions.
However, if I was new user, rather than waiting for me to discover what information was missing, the assistant should have proactively asked: "Which case type would you like to work with?"
Small prompts like that could make the experience much more approachable for someone learning Blueprint for the first time.
A Different Way of Thinking About Accessibility
I have used JAWS for a long time, and historically I have preferred direct navigation. It is what I know. It is what I trust. What surprised me during this evaluation was realizing how quickly that preference is changing.
The more I work with AI, the more comfortable I become relying on conversational interactions.
That does not mean traditional accessibility is no longer important. Users still need access to the interface and the ability to navigate independently. But AI introduces a new interaction model that complements traditional accessibility rather than replacing it.
For me, it often feels less like driving through the interface and more like collaborating with a knowledgeable assistant.
Looking Ahead
The experience also made me think about what AI could do next.
I would love to be able to say "undo the last change" and have the system automatically reverse the most recent action.
I would like AI to duplicate workflows, modify release planning information, and describe reports and charts in a meaningful text format.
As a blind user, one of the most exciting ideas is having AI help navigate the application itself. Imagine being able to say, "Take me to Release Planning" or "Take me to the next step," and being moved directly to the area where work needs to happen.
That would allow AI not only to perform tasks, but also to act as an intelligent guide through the Blueprint creation process.
Final Thoughts
When I started this evaluation, I expected to prefer the traditional Blueprint experience.
By the time I finished, I would have chosen the AI-assisted approach.
The reason is not simply efficiency. It is that the experience changed how I interacted with the application. The AI reduced navigation overhead, helped me learn as I worked, corrected small mistakes, and made the Blueprint process feel more conversational and approachable.
For a blind user who has spent years navigating software through a screen reader, that is a significant shift.
What began as an experiment in AI-assisted Blueprint creation ended up feeling like a glimpse into a different future for accessibility, one where understanding, navigation, and productivity can all happen through conversation.