Inspiration
Many people receive important letters from government offices, hospitals, and other organizations that are difficult to understand. These letters often contain complicated words, deadlines, and instructions.
This can be especially challenging for people who are not fluent in English, older adults, or people who have trouble reading.
I wanted to build something that makes these letters easier to understand so people know what they need to do.
What it does
DearClear is a web application that helps people understand complicated official letters.
Users can upload a PDF, take a picture of a letter, or paste the text directly into the website.
DearClear uses AI to turn the letter into a simple explanation. It also creates a checklist of actions the user needs to take and highlights important deadlines.
Users can choose different languages, including English, Spanish, Hindi, Tagalog, and Swahili.
The application also has a text-to-speech feature that reads the explanation aloud. This makes the information more accessible for people who prefer listening instead of reading.
DearClear also includes a demo mode where users can try a sample letter without needing an API key.
How I built it
I built DearClear using Next.js, React, TypeScript, and Tailwind CSS.
I used Google's Gemini API to process letters and generate simple explanations. The application is designed to identify important actions and deadlines from the original document.
I also integrated Puter.js for text-to-speech so users can listen to their explanations in different languages.
The website has a simple interface that allows users to upload documents, select a language, and view their results.
Challenges I faced
One of the biggest challenges was connecting the Gemini API. I ran into problems with model availability, API errors, and long response times.
I also had to make sure the application could handle different document formats and display the results clearly.
Another challenge was adding multilingual text-to-speech and making sure the audio controls worked properly.
To make the project easier to demonstrate, I added a demo mode that works even when the AI service is unavailable.
What I learned
This project helped me learn more about building applications with AI APIs and handling errors when external services are unavailable.
I also learned how important accessibility is when designing software.
Even a simple feature like reading text aloud or breaking instructions into a checklist can make a big difference for someone who struggles with complicated documents.
What's next for DearClear
In the future, I would like to improve the accuracy of document analysis and support more languages.
I also want to add deadline reminders, improve document scanning, and allow users to save their letters securely.
My goal is to make DearClear a useful tool that helps people understand important information and take action with confidence.
Built With
- google-gemini
- html
- javascript
- next.js
- puter.js
- react
- tailwind-css
- typescript
