Day 05 of 07
Iterate and direct a bigger build
Goal: Learn how to improve output with better feedback and how to break a larger project into smaller AI roles that make sense.
What to do
Give three concrete rounds of feedback
Ask for the three changes one by one or as a grouped list with clear confirmation before each change.
I am looking at the project you built yesterday. I want to make 3 improvements. Here is my feedback: 1. [Describe one thing you want changed], 2. [Describe a second thing], 3. [Describe a third thing]. For each change: confirm you understood what I asked, make the change, tell me if any change might break something else. After all 3 changes, show me a summary of what is different from the original.Review the updated version
Look at the new version and decide whether the improvements actually improved the product.
Break a bigger project into AI roles
Now practice bigger-project thinking: planner, builder, checker, and polisher.
I want to build something more complex: [describe a slightly bigger project]. Help me break this into a team of tasks: 1. The Planner - what needs to be mapped out first?, 2. The Builder - what needs to be built and in what order?, 3. The Checker - what could go wrong and how would we test it?, 4. The Polisher - what would make it look and feel professional? For each role, give me the specific prompt I should use to kick off that phase.Save the role prompts you will reuse
Keep the planner, builder, checker, and polisher prompts in your notes so you can reuse them on future projects.
Expected result
Your first project is better than yesterday, and you now know how to break a bigger build into specialist AI roles.
Key takeaway
- Good feedback is specific, testable, and limited in scope.
- Bigger projects need phases, not one giant prompt.
- Direction improves output more than volume does.