Thesys

AI research workspace for reading, comparing & citing papers

Productivity

Education

Artificial Intelligence

GitHub

Visit WebsiteSee on Product HuntGithub ⧉

The idea for Thesys started during my third year of university. I had a lot of reports and dissertation work to complete, and I found it really difficult to organize and work through academic the literature. I searched for tools that could help with academic research and came across Consensus. I used it for a while, but reached the end of the free tier eventually. That made me think about what I actually wanted from a research tool. I wanted something simpler where I could just upload a PDF, ask a question, and get an answer that shows exactly where the information came from in the paper itself. I didn't want to blindly trust an AI answer nor spend time searching through a paper to verify it. At the time, I was balancing university, personal responsibilities, and a part-time job, but I kept making time to work on the idea. After about a month, I had a working prototype. It wasn't exactly polished, and the UI and performance still needed a lot of work, but it worked. I started using that early version for my own research. It helped me gather and work through papers for my own reports and dissertation, I ended up achieving a First in both my reports and dissertation. That was the turning point for me because that's when I realized I was actually building something I genuinely wanted to use myself. From then, I wrote down the features I thought were needed to make it more functional and started working on them more seriously. The idea grew more than a PDF Q&A into features like summarization, bibliography exports, different reader modes, table detection, and eventually the to compare two papers around a specific claim. I also gave a quick demo to a course director at my university and she was genuinely impressed with how it worked and saw potential for it to help students to stay productive while working on large amounts of academic literature. Eventually, I open-sourced the project and shared it on Reddit, after which I shortly came to know that I wasn't the only person who faced this problem. I started receiving a lot of good feedback through comments. Researchers and students started using Thesys for their daily workflows, and many understood the value of having the source highlighted directly in the paper. Since then, the project has grown to around 150 GitHub stars and 20 forks, which has been really motivating. What started as something I built for myself has now turned into something other researchers are using for their daily work, and that is probably the most rewarding part of building Thesys so far.