You will practise turning vague requests into complete briefs and compare the results side by side.
Scroll back through your assistant's chat history and you will probably find a run of one-line requests: "summarise this", "make this more professional", "ideas for team bonding", "reply to this email". Each one got an answer, and you probably rewrote most of those answers yourself.
This exercise turns that history into evidence. You will take five prompts that were really used, rewrite each with the four parts from lesson 1.1, The assistant only knows what you put in front of it, run both versions and score them. By the end you will know which part of a brief matters most for the kind of work you do, which is more useful than any general rule.
Use prompts that were actually sent, by you or by colleagues who are happy to share them. Invented examples are too tidy. Real ones are short, a bit vague and written in a hurry, which is exactly what you are trying to fix.
Aim for a spread of tasks. A writing task, a summary, a request for ideas and a reply to someone are a good mix, because the four parts do not matter equally for each. If all five are emails, you will learn a lot about emails and not much else.
Copy each prompt word for word into a document, along with any text or file it was used on. You need the same input for both runs, or the comparison means nothing.
For each prompt, add context, a clear task, constraints and format, as covered in lessons 1.2 and 1.3, while keeping the rewrite as short as the job allows. A request to tidy up a two-line message does not need a paragraph of background, while a client proposal might need five sentences before the task.
Here is a worked example from Priya, the HR executive from lesson 1.2. One of her five was a prompt she had sent the week before: make this more professional, followed by a draft reply to a job applicant.
Her rewrite read: "I work in HR at a logistics company in Singapore. Below is my draft reply to a candidate who interviewed for a warehouse supervisor role and was not selected. He was polite and well qualified, and we may want to approach him for future openings. Revise the reply so it is courteous and clear without sounding like a form letter. Keep it under 120 words, keep my point about future openings, and do not add reasons for the decision that are not in my draft. Give me the email body only."
Context: who she is, who the candidate is and why the relationship matters. Task: revise. Constraints: length, what to keep, what not to invent. Format: email body only. It took her about two minutes to write.
Run the original and the rewrite in the same assistant and the same mode, each in a fresh chat so neither run colours the other. Memory or custom instructions, if you have them switched on, can stay as they are for both; module 6 deals with those.
Score each output from one to five for usability, meaning how close it is to something you would actually send or use.
1: unusable, you would start again yourself. 2: a few usable lines, mostly rewritten. 3: usable after substantial editing. 4: usable after light edits. 5: usable as it is.
Score the output, not the prompt, and judge each one before looking at its partner. Better still, ask a colleague to score them blind.
Priya's original produced a stiff, longer version of her draft that dropped the line about future openings. She gave it a 2. The rewrite kept the line, cut her draft by a third and read like a person had written it. She gave it a 4, because she still changed one phrase.
For each pair, write one line on which of the four parts changed the result most. This is the part of the exercise most people skip, and it is where the learning is.
Look for patterns across your five. For Priya, constraints made the biggest difference on the candidate reply, because "keep my point about future openings" stopped the model from dropping the most important sentence. Context mattered most on a request for team-bonding ideas, because once the model knew the team worked shifts across two depots, it stopped suggesting weekday dinners. Format made the biggest difference on a meeting summary, where asking for "decisions, then actions with owners" turned a page of prose into something she could forward.
As an example of how the numbers can come out, her five originals scored 2, 1, 3, 2 and 2, an average of 2.0, and her rewrites scored 4, 4, 5, 3 and 5, an average of 4.2. Yours will differ, and one of your rewrites may do no better than the original, which is worth recording too: it usually means the task was simple enough that a short prompt was fine, or that the missing piece was something none of the four parts covers, such as an example of good output, which module 2 deals with.
A finished table has one row per prompt and six columns: the original prompt, the rewrite, the score for each, and the one-line note on what made the difference. Hold on to the rewrites, since the best of them become the first entries in the prompt library you will build in module 9, and you will be surprised how often you reuse them.
Set aside about twenty-five minutes, open a document for your table, and start with the prompt from your history that annoyed you most.
Produce a before-and-after table of five prompts with scores and a one-line note on which part of the brief changed the result most.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).