← all ideas

AI Coding Tools and Methods

The idea

The argument is that using a combination of a planning AI and an AI coding tool can significantly streamline the development process, especially for non-technical individuals. Start by selecting a planning AI, such as Gemini 2.5 Pro, which can handle large amounts of context. This AI helps you outline your project and user flow by prompting you with questions and suggestions. For the coding tool, choose between ClogCode or Codex. Once you have these tools, create a master prompt that defines your role as a CTO and a prompt engineer, guiding the planning AI on how to interact with the coding assistant. If you have an existing codebase, prepare a project context markdown file and feed it to the planning AI. The AI will then generate a prompt for you to input into your coding tool, which will execute changes and report back. This method allows for continuous feedback and iteration, making the development process more accessible and efficient.

Why it works

• The argument is that using a planning AI like Gemini 2.5 Pro helps manage the complexity of large projects, ensuring that the context is well understood and the project goals are clear. By crafting a master prompt that defines the project scope and roles, you can guide the planning AI to generate detailed instructions for the coding tool, making the development process smoother and more efficient.
• One view holds that this method is particularly beneficial for non-technical individuals, as the planning AI acts as a bridge between the user's ideas and the technical execution, allowing for more intuitive and effective collaboration. This approach minimizes the gap between high-level thinking and detailed coding, ensuring that the final product aligns closely with the initial vision.

The playbook

1. Choose your planning AI and coding tool. For planning, select Gemini 2.5 Pro, and for coding, consider ClogCode or Codex.
2. Craft a master prompt. Define your role as a CTO and a prompt engineer to guide your planning AI on how to assist your coding process.
3. Create a project context file if you're working on an existing code base. This file will help set the context for your planning AI.
4. Engage with your planning AI. Start discussing your project ideas and user flow directly with your planning AI. This will help in generating initial prompts.
5. Follow the back-and-forth process. Copy the prompt created by your planning AI and paste it into your coding tool. Review the changes reported back, and continue refining your project with your planning AI.

Where people get it wrong

• Not Starting with a Clear Master Prompt: Many fail to create a detailed master prompt that sets the context for their project. Without a clear and structured prompt, the planning AI may not understand the nuances of the project, leading to misaligned outputs. To avoid this, start by defining the project's context, goals, and specific requirements in a comprehensive prompt.

• Ignoring the Role of the Planning AI: Some users jump straight into coding without leveraging the planning AI. This can result in inefficient workflows and a lack of strategic direction. Always use the planning AI to outline the project flow and feature requirements before diving into coding. This ensures that your coding efforts are aligned with the project’s goals.

• Failing to Iterate and Refine Prompts: Continuous improvement is key, but many overlook the importance of refining their prompts based on the output. The planning AI’s initial responses may not always be perfect; iteratively tweaking the prompts can lead to better outcomes. Regularly review and adjust your prompts to ensure they accurately reflect your needs and expectations.

Do this next

• Open an account with Gemini 2.5 Pro to use as your planning AI.
• Choose between ClogCode or Codex as your AI coding tool.
• Craft a master prompt for your project, detailing your role as a CTO and a prompt engineer.
• Start a new project by chatting with your planner AI about the user flow and project goals.