Communion

Communion Application's Logo

Inspiration

The app is based on most conventional social media app. The UI element is inspired primarily by Reddit

What it does

This application is designed to check whether the contents of the user's post are true or not by using Claude Sonnet-4.6 to web-scrape and summarize that. However, before we let our user post anything, the AI will scan if the content within the post follows the laws and the app's regulations or not. By following this rubric, SCORING RUBRIC (per claim, 0-100) 0-29 contradicted by sources, internally inconsistent, or extraordinary claim with zero support 30-49 unverified or only weakly supported (single low-authority hint) 50-69 partially supported or limited corroboration 70-89 well-supported by reliable sources 90-100 confirmed by multiple authoritative sources After AI has verified, the post will be visible on the platform, and people can comment on and rate it, which is the part where humans become an important point of the application. Some of the posts cannot be summarized by an AI. For example, if the light just went out in a city, will AI immediately know? Of course not. So it is going to be person-to-person to pass the information to each other. We know that AI can make mistakes, so we added a separate user rating system along with a comment system. additionally, the AI will only use information from trusted, copyright free source

How we built it

We started by brainstorming and planning our roles in this project. Including a logo designer, a software designer, 2 UX/UI designers, and an AI Engineer. In the first 3 days, we start by designing our application's UX and UI design. On the 4th day after we start, we use Claude Code + Claude Design to implement our designs, functions, and ideas into the application, then troubleshoot the bugs in Claude and prompt it to fix and improve the application. Surprisingly, we did make our core function work, but it is not truly completed. On the 5th day, we improve and add more features to our application to make it look more alive (including artificial posts for a look) and go live on Render.

Challenges we ran into

The time we have is limited, so we must manage the time wisely. Our resources are limited. Every time we use it, we must think twice before executing anything.

Accomplishments that we're proud of

Most of the applications are working as we expected.

What we learned

We learn how to work as a team with direction. We learn how to manage a project and AI implementation in our workflow. We get to share our thoughts in this project and look at many perspectives.