DIZ-AI

Inspiration

I'm a high school student, and I've seen firsthand the devastation that natural disasters can cause. Even though I live in Utah, the ripple effects of the California wildfires impacted my community, making me realize how unprepared most people—including myself—really are. When disaster strikes, how do you know what to do? How do you prepare when there are endless options and advice? I wanted to simplify this process and create something that helps people get ready quickly and effectively.

What It Does

My app uses the Wolfram API, OpenAI’s API, and some creative backend work to generate a personalized natural disaster preparedness plan. By analyzing a user’s location and unique situation, it creates a customized checklist of what they need to buy, what steps they should take, and which disasters are most likely to affect them. It also provides action plans based on the specific threats in their area, so they know exactly what to do if an emergency happens.

How I Built It

As a new solo developer, I used every tool I could find to make this work. I started with paper flowcharts, turned them into structured prompts for ChatGPT, and then used AI-assisted coding to bring my ideas to life. For the front end, I experimented with V.0 to design a simple and functional interface.

To be specific, I got the idea, then went to youtube to find out how to use API keys. I saw that wolfram was an option and I tried to utilize it and that was how I really learned how to use API. I then found out that you can get Openai API easy, and then spent the rest of the time prompting and trying to make a functioning logic. Much of this process was screenshotting the local host and then sending it back with lots of directions, just to see the ai get it wrong again. But by doing this process I weeded out some of the errors and got the sourcecode to work.

Challenges I Faced

Going into this project, I was pretty naïve about the challenges of coding with AI. I quickly realized that fixing one issue often created a whole bunch of new ones, leading to an endless cycle of prompting, debugging, and re-prompting. Some of the biggest hurdles I faced included:

Connecting APIs and getting Wolfram to work smoothly with OpenAI Formatting the data returned from OpenAI in a way that made sense Figuring out how to submit the frickin data to an API properly FORMAT OF CODE. So confusing to me. It felt impossible at times, but I pushed through and learned a crap ton

Accomplishments I'm Proud Of

This is my first project, and the fact that it actually works is something I’m incredibly proud of. Before this, I had never used an API, and honestly, I had barely coded at all. But I built something that can genuinely help people, and that makes me hella excited for what im gonna do in the future.

What I Learned

Being a solo full-stack developer when you don’t know how to code is... difficult. But I also learned that: Structuring AI prompts properly makes a huge difference Keeping track of previous versions is essential (because sometimes AI messes things up) Problem-solving is 90% of coding, and persistence is key

What's Next for DIZ-AI

I want to make the responses more personalized and faster. Adding user accounts and data storage (so people can save their plans) is also a top priority. Another big goal is to include real-time updates and make the UI more accessible and user-friendly.

This is just the beginning, and I’m excited to see where it goes next, and even more excited to see where I go next.

Built With

• css
• html
• javascript
• node.js
• react.jsopenai-api
• tailwindcss