One of the most effective AI productivity hacks I've adopted recently is using a single file logic until a backend or complex logic becomes absolutely necessary. This approach varies with each project, but the core principle is to minimize unnecessary components until you hit a limitation.

Inspired by @levelsio's coding style—where entire production applications are built within a single file—I've found that most of the time, you don't need a web app, backend, or even React. Agents can assist in writing the file and filling in any gaps you may encounter.

In many scenarios, a simple HTML/JavaScript file suffices, utilizing localStorage or a single Python structure to simulate logic and storage. This method helps maintain focus on delivering immediate value and facilitates easier iteration.

Sharing the project is also straightforward; you can effortlessly send it via Teams or Slack without complications. Embracing minimalism, simplicity now often outweighs the complexity of future solutions.

Dictation Over Typing

The second approach is (obviously) to replace typing with dictation. Since dictation is faster and much of my coding has shifted to prompting and reviewing, using Yakki to dictate while working makes a lot of sense. Dogfooding is a powerful concept.

AI as a Senior Partner

Last but not least, I am leaning on AI, using it as a senior partner. For every operation or meaningful task, I allocate some time to consult with AI and get a second opinion on what I am going to do. Asking to identify gaps and to provide alternative or potential improvements. This is one of the biggest drivers of not-so-obvious value I have added lately.

What are yours?