Set up the system before you ask it to build.

Create a useful AI workspace.

Start with a small, focused workspace. The goal is not to build a perfect system. The goal is to make it easy for an assistant to understand your work and continue from the last decision.

• Create one folder for your project and keep decisions, notes, and drafts there.
• Add a short project README with the purpose, audience, and current status.
• Keep one place for prompts that worked well so you can reuse them.

Prompt your dream project clearly.

A good first prompt gives direction without pretending the final answer is already known. Ask the assistant to question weak assumptions before it starts producing assets.

Please help me shape a practical AI project.

Context:
- My audience is:
- The problem I want to solve is:
- The result should help people:

Before proposing a solution, ask me the 5 questions that would most improve the project direction. Then suggest a first version I could build in one week.

Write agent instructions once.

An instruction file keeps your preferences out of every prompt. It should explain the project, your standards, how decisions should be made, and what must never be changed casually.

Create a concise project instruction file for this workspace.

Include:
- What this project is
- The files and folders that matter
- How to run and verify it locally
- Design and content standards
- Things an AI assistant must not change without asking

Ask me for missing details before writing the final file.

Use voice capture for rough thinking.

Voice is best for messy context: what happened, why it matters, what you want next, and what feels unclear. Then ask the assistant to turn that context into a structured brief.