I have enjoyed creating different kind of content for as long as I know. It started with scribbles then jumping to drawing, photography, video and writing. I m not good at all of them but I have given it a shot, some more than others.
I’ve also experimented with development which helped me bring logic and rationale to the fore front of my thinking process. It also helped me understand the practicality of ideas that could otherwise look only very tempting.
More recently though I have been having an itch to start writing again. A bit more niche than what I used to write. I do have insights to provide after all I have been in the industry close to 20 years now.
However given that development has felt alien to me for a long time, I haven’t really touched a line of code in about 15 years and I didn’t really want to start with a template to start a blog (designer by heart, won’t use a template).
Yes I am aware I should test waters with the bare minimum and see if it gets traction, this is where it gets interesting. Vibe and agentic coding models have been around and I’ve been wanting to play around with them for a while, this was the perfect excuse to start.
This aligned pretty well with the holy month of Ramadan, for those who don’t know it’s a month of purification for us Muslims (What’s Ramadan?). Living in a Muslim country this means our work hours are shorter given that most of us if not all are fasting from dawn to dusk.
I m not like an AI virgin but I hadn’t played with the agentic coding tools, so.. One of the things on the list was to play around with the no code AI tools or more like conversational tools that would help me build my blog, give me more insight and help me grow.
Claude had been pushing out amazing releases which got me interested. I must say I did experiment with Bolt before with Figma import but you can only do so much on their free plan that I left that exercise shortly. I wanted to try Lovable as well but never got around to it.
I had a discussion with one of my best friends who happens to be an amazing dev, brainiac of a person and an amazing soul, Ahmer Sultan (cyphorous). Instead of pushing me to go with Claude AI right from the start he let me explore a few (TRAE, Cursor, Replit) then guided me towards Cursor (since I did know the basics of development – enough to call bullshit when working in teams).

Getting into anything newer to oneself, you need to afford the time to actually understand it, have a purpose and have the completion mindset in respect to what you’re learning.
In this case since I had set my sights on Cursor (more like guided towards it) I asked Ahmer to help me get to speed which he did, walking me through his process, explaining the planning stage, agent selection so on and so forth.
Getting on Cursor is not really hard but its better if you have some technical knowledge. Even if you don’t, doesn’t really take too much to get up to speed. If you’re unsure of something it will answer you or guide you through the process, it has these different modes which obviously have different purposes they do help you get things done faster.

Now I already had a design in place which I wanted to convert to being usable. The way I started with it was to make sure I used exported images from Figma for all the pages, in addition to exporting some components separately which needed more context. Then I exported the images I needed to be used with in the output including svgs & pngs.
For some of the components I would go in more detail depending on behaviour, and if it had to be different based on screens (viewports). This was important to make sure everything worked as intended and didn’t break on on smaller screens (I do think some improvement is needed). For some pages I would use cross context (explain similar functionality on other pages and ask it to replicate with changes) and some were no brainers.




Some components it would add additional decorations (trying to be helpful?) but I had to tell it to calm down and focus on my words 😉 .
There are a few models you can use to get into it. I ended up using Chat GPT Codex for planning and breaking things down into workable items. Then setting it on Auto for the rest.
One of the things I learned is instead of giving big chunks of tasks, work with small ones so it’s easier to fine tune it. I did this with the initial setup for WordPress and it worked great. I implemented the same method in most places, except where the context needed to be explained in detail.
The result of this was this blog, now I know there might be some bugs here and there and I’ll squash them as I go by. I’ll be happy to answer any questions, drop in a comment here or drop me a line via Reach me.
This blog is relatively new so if you come across something awry, point it out. 🙂
OH NO!
People aren't going
gaga here
Be the First one 😉