Starting Again
The last few months have been a bit of a roller coaster. I was made redundant in June 2026. The job market was a bit brutal. I was lucky to have a role lined up soon after, but the contract put me off a bit. So I decided to wait, which extended the job search by another 4 weeks.
There are plenty of really good candidates because of mass layoffs. And the interviews are, well, quite messy too. Most companies are still figuring out how to interview in the AI era. That makes preparing for an interview super confusing. I never know if I should prep with AI tools or without them. This period of job search and uncertainty was not playing well for me mentally, so I wasn’t in a good space to produce content. Towards the end of July I managed to secure a role in the construction software space, which is a completely new domain for me and a completely fresh stack, given everything is on .NET Core and Oracle.
Why am I changing my approach to content
With the rise of AI tools, everywhere I look I see so much AI-based content. Sometimes the repetitive nature of it makes it obvious that it’s basically an AI dump. I’ve used AI to write and edit almost all of my previous articles. However, I’ve now started to appreciate that content we create that is quite personal should not be outsourced to an AI agent. I loved LLMs for this, mostly because English is my second language. My grammar and correctness will always have problems, and AI would make it sound clean and professional. But AI agents go the extra mile and change almost everything, to the point where it doesn’t feel authentic anymore. So now, other than Grammarly or a spellchecker, I don’t think I’ll be using AI in my future content.
What am I going to use AI for in the future
AI has become a fixed part of my workflow. Whether I’m using Pi + DeepSeek or Claude Code, I don’t think I’ll be typing code anymore. I’ll spend a lot more time thinking and coming up with experiments, documenting how certain architectures and patterns should look so the agents can work within the constraints and produce code the way I would write it, or the way I would train a junior engineer to write it.
AI will always be a source of research and vibe coding. Sometimes I need to get an idea out of my head, and vibing it helps me figure out if it’s even viable. So spending an hour hacking a prototype and using that spike to decide whether I want to go ahead with the idea isn’t going anywhere, IMO.
What I’m after this time
So this time, I will continue to write about my experiments, the projects, and how they are going. However, it will come with a decent number of spelling and grammar mistakes. Especially when I’m too lazy to copy the markdown into Grammarly to check.
