Build in small loops with clear review points.

Continue building your big thing by making it smaller.

Ask for the smallest useful version that proves the idea. The first version should be narrow enough to finish, but real enough to teach you something.

• Define one audience and one outcome.
• Remove features that do not prove the main value.
• Ask for a first version that can be reviewed in one sitting.

Prompt a personal assistant for the project.

Treat the assistant as a small team. Assign a role for the next task, then ask it to produce work and critique the result.

• Researcher: find options, risks, examples, and missing questions.
• Builder: create the first version with simple, direct implementation choices.
• Reviewer: look for bugs, unclear copy, accessibility gaps, and weak assumptions.

Act as a practical project assistant.

For this task, use three passes:
1. Research the options and risks.
2. Propose the simplest implementation path.
3. Review your own plan and identify what could fail.

Keep the output specific and actionable.

Turn image or design ideas into a usable brief.

When you have a visual direction, do not stop at mood. Translate it into layout, hierarchy, components, states, and constraints.

Choose the best stack for the job.

The right stack is usually the one that lets you verify the idea with the least unnecessary machinery. Ask for tradeoffs, then pick based on deployment, maintenance, and your own skill level.

I want to build this project:
[describe project]

Recommend a simple technical stack. Compare 3 options using:
- time to first version
- deployment complexity
- future flexibility
- what can go wrong

End with one recommendation and explain what you would not build yet.

Create a plan that can survive real work.

A good plan includes what to build, what to skip, how to verify progress, and when to stop expanding. Keep the first milestone close.