My first attempt working with Claude

My first impressions of working with Claude

Like many of us in the software development space, the use of A.I. has become an unavoidable situation. A lot of debate is going around about the future and if A.I. will some day take our jobs or not.

My personal opinion is that A.I. is another tool that will boost our productivity, but never our creativity or judgment, at least in our career since other fields like arts has impacted a lot.

Enough with the intro about the usage of A.I. as I mentioned above I started working more cautious with A.I. both for daily use and for my job. I worked with ChatGPT and Copilot mostly for fixing small things or formatting my code, of course with the old school way of (Ctrl+C, Ctrl+V). At this point A.I. for me was something like an enhanced Google/Stack Overflow.

And then it came, my dream A.I. tool:

Cursor IDE

This bad boy could do anything and the best part is, it does it on a VS code clone IDE the one I am so used to work with that felt like the perfect tool for me as a developer. At this stage I was also more deep into A.I. in general, I was learning about MCPs, how to write better prompts (it’s a skill trust me), the difference between the models and their “powers”, hooks, Skills and all the fun stuff. I got so used to working with Cursor that to this day I don’t know if I will feel the same using another platform, seeing the work being executed inside the IDE, having control and guiding the model, everything was perfect!

So why did you switch to Claude?

Honestly because I wanted to see how much better this Code Genius is, the one that was so “dangerous” they had to hide one of its models from public eye. So yes, curiosity won…

But is Claude Code actually better?

To put it simple, yes it is better in terms of how clever it feels. I was proposed solutions that other models never suggested or even if I came up with the idea they were unable to implement them without having to debug for hours and hours. Currently I am using Opus 5.5, which as I write this blog is the latest Opus; it is truly unique in its way to search for solutions that with previous models I had a very hard time. Actually the whole 3D live Atlas mode in my website (geotsinis.com) was revamped using Claude and boy it made a huge difference, especially with the whole Expedition mode and the animations. My friend Grok, although still new to the whole coding world, honestly felt like a downgrade as a main agent on its first release. So Claude, at least until now looks like miles away in terms of power and skill.

Any drawbacks?

As I said I am still familiarizing myself with the new platform, I am using the Claude.ai interface to write my prompts and organise my projects, because I can’t bring myself to enjoy working with the terminal, I’m sorry, I need to see an interface to help put my thoughts into the right order. But even the Claude.ai app doesn’t replace the feeling with working on a clever IDE it’s like having a chatbot app which can open terminals and diff files, so this needs enhancement for my tastes. Another huge difference I noticed was that with Claude I am consistently checking on my usage meter, I made the mistake to use it like Cursor in the beginning, writing requests one after the other in the same session and my tokens were flying away faster than your bank account balance on the first day of the month. I then learned the best practices for spending less tokens and being more efficient with using Opus and I believe I have a more balanced experience, although I felt so traumatized after seeing my 5-hour limit vanishing after just a few prompts, that I still keep checking on the usage all the time, something that was never an issue for me using Cursor.

Another difference I noticed right away, was that the agents are acting more autonomous, I don’t know how to explain but with Cursor I felt like I had a lot more control of the process. Claude sometimes feels “too clever” and it might feel less procedural, which for some people will sound more like a pro instead of a negative.

To summarize

Will I keep using Claude after the first months?

Will I run back to Cursor if it makes a comeback?

Will I get rich using A.I. and stop working, or will I just use it to ask for survival tips in the wild?

We will find out. This humanity’s technological leap still feels new to a lot of people and I believe we are barely scratching the surface yet, since it also evolves exponentially. For once, I feel excitement with all this new era, learning so many new things that I never even imagined when I was still a student in my university. I believe it’s a huge chance for people enjoying learning and discovering new technologies. Technologies that we might have actively avoided in the past because of the learning barrier or the resource shortage.

Until next time, I am George Tsinis and you’ve been reading my first article.