Start with one question
A first project works best when its purpose fits in a sentence. Instead of “build an app,” try “help someone compare three study spaces.” Name the person, the decision they need to make and what a useful answer would look like. A small scope gives you something you can finish and examine.
Separate learning from polishing
List the skill you want to practice and the smallest output that uses it. A paper interface can teach you about interaction. A short script can teach you about input and validation. A spreadsheet can show whether a calculation answers your question. Use a format you can inspect instead of choosing a tool only because it looks impressive.
Make the first version testable
Write two or three checks before you begin: can someone understand the label, can the calculation handle a blank value, can the page be used on a small screen? Keep a record of what you expected and what happened. A useful failure tells you where the next version should change.
Show the decisions
Explain your starting question, your choices and a limitation. Avoid describing an unfinished prototype as a complete service. If you use made-up data, label it. If an AI tool helps you, inspect its output and be ready to explain the parts you keep.
Reflect before adding features
Ask what became clearer, what still confused someone and what one improvement matters most. This Discover → Develop → Demonstrate → Evolve rhythm keeps the project connected to learning. Try a self-guided studio challenge for a starting brief; it is a practice activity, not an enrollment or certificate.