I remember the first time I used Claude. Unfortunately, I didn't find a way to share the chat to prove my claims. It was during a take-home exam, and we were encouraged to use LLMs, as long as we disclosed it. I asked the same questions to ChatGPT, and I found it was much better. Now, I know some people might say that's cheating, but I wasn't cheating—I needed help. Writing essays has always been a struggle for me.
My difficulties with writing began in high school. Before that, it wasn't as challenging, as I was mostly writing novels and sci-fi, genres I was fairly decent at. But academic essays were a different story. After my first attempt, I decided Claude wasn't worth my time. That was almost a year ago.
In June, I heard about the release of Claude 3.5 Sonnet and saw the great reviews. At first, I thought it was just overhyped, but I still decided to give it another shot. It was during a time when ChatGPT was starting to feel outdated, struggling to keep up with new tools and libraries in the web development world. As you may know, the web dev environment is constantly evolving, with new technologies and frameworks emerging all the time.
Claude Sonnet, however, was a game changer. It was incredibly effective for web development, which is primarily what I use LLMs for when coding. My experience with it made me think that there should be a new benchmark for web development (and interface design) involving LLMs, because it made such a noticeable difference.
A few months before trying Claude, I had experimented with Cursor using ChatGPT. The experience, while useful, wasn't life-changing. However, when I tried Claude, everything changed—it pushed Cursor to another level, making it far more impactful for my workflow.