Prompting and Collaboration
Build a prompt from its four parts, give the model what it needs, ask for the output you want, and improve a result instead of starting over.
- Lessons
- 7
- Exercises
- 39
- Minutes
- 41
- 1
Anatomy of a Prompt
After this lesson you can take a vague request apart into task, context, constraints and format, and see which part a weak prompt is missing.
Code orderFill blankMultiple choice - 2
Giving Context
After this lesson you can tell what a model cannot know about your situation, and hand it that instead of hoping.
DebugMultiple choiceTap - 3
Specifying the Output
After this lesson you can ask for the shape, size and audience of an answer, so what comes back is ready to use instead of ready to edit.
DebugFill blankMultiple choice - 4
Iterate, Don't Restart
After this lesson you can turn a nearly-right answer into the right one in one or two turns, instead of rewriting the prompt from scratch.
DebugCode orderMultiple choice - 5
Deciding What to Delegate
After this lesson you can split a piece of work into the parts a model should do, the parts you should do, and the parts you should do together.
Code orderMultiple choiceTap - 6
Checkpoint: Building the AskCheckpoint
Take prompts apart, add what is missing, and decide what to hand over.
DebugFill blankMultiple choice - 7
Boss: Fix a Failing PromptBoss
One vague request, improved through eight decisions until the answer is ready to send.
DebugFill blankMultiple choiceTap