Truthly AI

False case
true case
Sources

Truthly

A Simple Way to Cut Through the Noise

How It Started

Truthly came together during the 8th edition of MariHacks, our school’s hackathon at Marianopolis College. We were huddled around a whiteboard, bouncing ideas around, and one issue kept coming up: misinformation. It’s hard to avoid, especially during elections or big news events. We wanted to build something lightweight and useful that could help people check what they’re reading without needing to be an expert.

What Truthly Does

Truthly is a Chrome extension and website that helps users fact-check online claims in real time. Highlight a statement, and Truthly uses the Perplexity API to verify it and show supporting sources. It’s fast, simple, and gives people a way to question what they see instead of just scrolling past.

Building It in 24 Hours

Time was tight, so we had to be smart about how we worked. We used Cursor AI, an AI coding assistant, to help us quickly scaffold the framework of our code. It saved us hours and let us focus on building features that mattered. It was a reminder of how AI isn’t just about flashy tools, but about democratizing coding, helping more people bring ideas to life, even under pressure.

We built a clean Chrome extension that lets users highlight text and instantly get a summary with linked sources. The web version offers a deeper dive. We had to learn how to build extensions, engineer smart prompts, and handle messy or vague input.

What We Learned

We realized that fact-checking isn’t black and white. Sometimes there’s no simple true or false. Validity often depends on how something is worded, what the context is, and where the info comes from. This was also our first time using prompt engineering for the Sonar model. We saw firsthand how minute changes in wording drastically affected the API results. We also sharpened our full-stack skills, learned how to prioritize under pressure, and thought a lot about the ethics of working with real-world data.

The Tough Stuff

The biggest challenge was time. We had to cut down features and focus on getting the core experience right. Dealing with ambiguous claims, API limits, and keeping the UI intuitive pushed us both technically and creatively.

What’s Next

Truthly started as a hackathon project, but we see real potential in it. We want to improve how it interprets claims, expand our verification tools, and build for more platforms beyond Chrome. Most of all, we want to keep creating tools that make truth easier to find, and show how AI can help anyone build things that matter.