Prompt Engineering Essentials · v0.5

Shape complex work and improve it

Module 4 of Prompt Engineering Essentials: shape complex work and improve it.

Course contents and progress

Course progress

0 of 10 course sections complete

Eight modules, final knowledge check, and capstone.

This course remembers completed sections and your last course page in this browser on this device. RayAI does not receive this progress or anything you write. Unfinished answers and written responses are not saved, so an interrupted activity restarts when you return.

MODULE 4 | 40 minutes | FAST TRACK CORE + GO DEEPER

Shape complex work and improve it

Core idea: complexity needs structure, not incantations

When a task contains several different kinds of work, separate them into stages. Each stage should produce something a person can inspect before the next stage begins. One prompt can contain multiple steps, but one primary deliverable remains a strong default.

  1. Clarify: ask concise questions when a missing input would materially change the answer.

  2. Analyze: extract facts, constraints, themes, or options before writing a recommendation.

  3. Draft: create the requested artifact using the approved evidence and structure.

  4. Review: test the draft against explicit success criteria.

  5. Revise: change the smallest instruction that addresses the diagnosed problem.

Use output structure intentionally

Specify a list, table, outline, email, checklist, schema, or narrative when the form matters. Do not request a table simply because it looks organized; use it when rows share comparable fields.

COMPARISON TEMPLATE

Ask for inspectable work, not hidden reasoning

Different model families handle reasoning prompts differently. Asking a model to reveal hidden chain-of-thought is not necessary and may not improve performance. Ask instead for a concise rationale, material assumptions, evidence used, visible calculations, alternatives considered, and unresolved uncertainty.

See it: a staged event workflow

STAGE 1 - EXTRACT

STAGE 2 - PLAN

STAGE 3 - REVIEW

STAGE 4 - REVISE

Improve one dimension at a time

Observed failure Targeted revision
Too generic Add the missing audience, decision, or source facts.
Wrong structure Specify fields or show one correct example.
Invented information Constrain the source and define missing-value behavior.
Too long Set a length, priority order, and what to omit.
Ignored a constraint Move the constraint next to the task and test it with a representative case.

Try it: practice the skill

What to do

  1. Choose the task suggested for this module, or use another simple, low-risk task.
  2. Open ChatGPT or another AI assistant in a separate tab or window.
  3. Enter your prompt and send it. Do not include private or sensitive information.
  4. Read the AI answer.
  5. For this module, identify one problem in the answer. Change one instruction to address that problem, then run the revised prompt.
  6. Return to this course and complete the short notes below.
    If you cannot use an AI assistant now, read the worked example and write what you would try later.
Path Your task
Everyday life Split your comparison into extraction, comparison, and decision-support stages.
Learning Create a teach-practice-quiz loop and require feedback on the learner's answer.
Work Separate fictional note extraction, email drafting, and final approval into distinct steps.

Module 4 test - check before you continue

Answer Questions 1–3 by choosing A, B, or C. For Question 4, write a short response. After you finish all four, use the answer key at the end of the course to check your work.

Your response stays in this page and is not saved or sent to RayAI.

1. A task contains research, comparison, drafting, and final approval. What is the strongest approach?
Review after attempting

Best answer: B. Separating research, comparison, drafting, and approval creates inspectable checkpoints and keeps final authority with a person.

2. Which is the better positive instruction?
Review after attempting

Best answer: B. A positive instruction describes the desired evidence and structure. 'Do not be vague' does not define what useful specificity looks like.

3. What should you request instead of hidden chain-of-thought?
Review after attempting

Best answer: A. Inspectable rationale, assumptions, evidence, calculations, and uncertainty support review without demanding hidden chain-of-thought.

4. Diagnose one weakness in your latest output and write one targeted revision instruction that addresses only that weakness.

Review after attempting

Your answer should you name one observed failure and write a specific change tied to that failure without rewriting unrelated instructions.

Finish this module

Mark this module complete when you have attempted every activity you intend to complete and reviewed your work against the v0.5 guidance.