After having spent over a decade in jumping onto solving problems, I think it is time to give this idea a break. The idea to to not jump onto solving the problem but to first understand the problem. Old habits are difficult to break but it is time to get tough on self.

I’d try to add atleast 30 problems per week but no promises to self. These problems can be from existing projects or new things that I am building. After writing about it, I’d spend 1 hour to explain each problem and think about the possible solutions. Then once everything is planned, I’d spend exactly 1 hour to solve the problem. This way I know I am not wasting time and spending enough time rather than manual context switching. I would also refrain from trying to solve the problem in the same day. It would be best to solve it the next day or mark it in the calendar. This page is where the problem statements will be documented. Everything will be done in public. The idea is to display the process and the thought process. AI can code better, so there’s no hiding.

While writing about the problems, I’d be using multiple AI models to do the thinking. This way the headvy lifting is done by the AI and I can focus on architecting the solution. Also, I would be attaching the prompts and response from each model. This way, I’d know which model is better in terms of cost, speed and quality.

Current Models

• Claude 3.7 Sonnet
• OpenAI GPT-o1
• xAI Grok