As LLMs continue to evolve, I’ve found myself changing the way I write prompts. I’m curious have you changed your prompting style over the past year? What techniques or habits have made the biggest difference for you?
From my experience since the early ChatGPT era (around 2023), prompting has become much simpler. Early on, people relied heavily on structured “prompt engineering” with templates and techniques, but newer LLMs understand natural language much better, so clear and direct instructions are often enough.
That said, it hasn’t become useless—just less rigid. You still need to be precise about what you want, especially for complex tasks, but it now feels more like communicating with a capable assistant than carefully engineering every prompt.
For me, the biggest lesson is that AI can help close a lot of technical gaps, but the real difference between people is still creativity: tools can help you build faster, but finding an original, useful idea is much harder.
One of my most valuable projects didn’t work out as planned, but it taught me the importance of data quality and clear problem definition. Those lessons have been more useful than any successful demo.
Any Model that you are using right now, if you want to optimize the result just ask the model itself of how should I prompt you, It will provide best techniques out there to get the most out of the model
In my case, giving persona to the model and asking it “iteratively work on a task” could help you get most out of the model