Introduction
Recently, I switched my main AI tool from ChatGPT to Claude (though to be precise, I’m still using both for now).
I had been using Claude here and there as more of a secondary tool for some time, but ChatGPT had always been my main one. Since I had already been paying for ChatGPT, I felt reluctant to switch and had a hard time committing to using Claude seriously. But once I decided to make the move, I noticed quite a few things, so I wanted to share them here as a personal note.
Why I Hesitated, and Why I Finally Switched
I was generally satisfied with ChatGPT, but I sometimes felt it was a bit too agreeable, which made it slightly less satisfying for brainstorming or using as a sounding board for critical thinking. Some of that could be improved with better prompting, but there had long been a sense that the overall feel just didn’t quite match what I wanted.
On top of that, I hadn’t felt much momentum in my work over the past year, so I wanted some kind of change. Around the same time, I also started hearing more and more people talk about moving from ChatGPT to Claude, and that made me think I should seriously give it a try myself.
That said, because all of these tools fall into the same general category of AI chatbots, and because I had already tried Claude, Perplexity, Gemini, and others while still keeping ChatGPT as my main tool, I had thought of them as more or less “similar enough.” That made it surprisingly hard to fully commit to switching. More specifically, I had concerns like these:
I had work-related tools built with GPTs — ChatGPT has GPTs, which users can customize and create. They’re useful because you can make them for your own personal workflows as well. However, creating and editing them requires a paid plan. So if I ended up switching to Claude as my main tool and moving ChatGPT to the free plan, I was concerned that maintaining and updating the GPTs I had made for myself might become difficult.
I wasn’t sure whether the answer quality would meet my expectations — As mentioned earlier, I did feel a certain lack in ChatGPT’s responses, but there were no major or fatal problems with it. I had heard good things about Claude, but switching a tool that I rely on heavily for work is still risky. Even if it might improve over time, it could just as easily get worse.
In the end, when your dissatisfaction isn’t that big to begin with, changing tools feels heavier than you’d expect.
What I Learned After Actually Using It
Once I started using Claude seriously, both of those concerns turned out to be resolved much more easily than I expected. For my own use cases, Projects could mostly replace GPTs, and the response quality was more than good enough for me. In fact, I even had the impression that Projects allowed more detailed customization and felt more flexible than the GPTs I had been using.
What struck me even more, though, was how different the actual experience felt, even though both are AI tools in the same category.
My impression was that ChatGPT tended to be more agreeable and often built on whatever direction I was already leaning toward, whereas Claude felt more critical and more willing to make firm judgments. But it didn’t feel like it was making those judgments carelessly — rather, it felt like those answers came after some actual thought. That suited me better.
I had also been somewhat dissatisfied for a while with the way ChatGPT sometimes tried to produce an answer even when my information was vague or incomplete. Claude, by contrast, seems much more likely to hold back from giving a definite answer when it judges the information to be insufficient, and instead asks follow-up questions first to draw out what it needs. Honestly, that was eye-opening for me. Which style is better probably depends on the user and how they use the tool, but for me, Claude feels like the better fit.
Overall, Claude gives me the impression of being more interactive and better at leading the conversation. Even when I can’t clearly explain what I want, it feels like it helps me organize a path toward the thing I’m actually looking for. I find that very comfortable.
That said, I should note that I have not done a careful side-by-side comparison by asking both tools the exact same questions. These are simply my subjective impressions based on having started using Claude more seriously.
That Said, I Still Don’t Know Yet
Even so, it hasn’t been that long since I started using it, and there are still many things I haven’t fully tested.
Just in case, I’m still keeping my paid ChatGPT plan for now and using both at the same time, so I haven’t yet reached a final conclusion about which one is better for me overall. Depending on how things go, I’m even considering keeping paid plans for both, though that admittedly feels a bit wasteful.
Since I decided to make the switch, Claude released Opus 4.7 with benchmark results that drew a lot of attention, while ChatGPT has also continued rolling out upgrades and releases at a very fast pace, including GPT-5.4, ChatGPT Images 2, and GPT-5.5. Both companies seem to be evolving rapidly. It’s extremely difficult to decide which one is truly better, and even if one seems to pull ahead for a while, it feels like the other may quickly catch up and overtake it again.
At least for me, both leave a good impression right now — but I’m not overly optimistic either.
Keeping at It Comes Before Mastering It
What this experience of switching tools reminded me of is that until you actually start using a tool, you can’t even tell whether it suits you or not.
Before worrying about whether you can master a tool, what matters first is starting to use it — and continuing to use it. That requires making the decision to simply take the first step. And once again, I felt that taking that first step was the hardest part.
To be honest, I had been interested in moving to Claude for quite a long time. But for the reasons I mentioned above, I couldn’t quite make the decision. This time, though, I finally went ahead with it, and it reminded me once again that you really don’t know until you try. I still don’t know whether the outcome will ultimately be good or bad, but at the very least, most of the worries I had before starting turned out not to be such a big deal in reality.
Not just with tools, but in general, I wonder if many of the things that make us hesitate come from the fears and assumptions we create before we even begin.
Conclusion
The real nature of a tool rarely reveals itself right away. You usually can’t understand it by just trying it briefly. It’s only by continuing to use it that you gradually begin to see your compatibility with it, as well as the particular habits or quirks the tool itself has. I think that’s just how it is.
What this transition reminded me of is that in order to understand a tool, “continuing to use it” comes before “mastering it.” In my case, I spent a lot of time hesitating because of all the concerns I had before even starting. But rather than trying to judge too quickly whether something is good or bad, it may be more important to spend a certain amount of time with it first. That mindset itself may be an important part of choosing the right tools.
Thank you for reading!
Top comments (0)