Karim, the production manager at Delta Garments, does not use an AI tool for every task, and he should not: for some tasks, writing it himself is faster and safer than describing it to a tool, checking the result and fixing it. For other tasks, an AI-drafted first attempt genuinely saves time, because the checking is quick and the writing was slow. This lesson gives Karim, and you, a simple way to decide which is which before starting the task, instead of finding out halfway through.
The decision is not about whether AI tools are generally good or bad. It is about matching the tool to the task: how well-defined the task is, how much checking it will need either way, and how much of the value is in the wording versus in the judgement behind it.
You need: the habit of writing a clear instruction (Lesson 2.1) and reviewing an AI draft for substance (Lesson 2.2).
By the end of this lesson you can:
Not covered here: which specific AI product to use, or a general claim about how much time AI tools save. This course teaches a decision method, not a fixed rule for every task.
| Term | Meaning |
|---|---|
| First draft | The first attempt at a piece of writing, produced either by a person or by an AI tool, meant to be checked and improved. |
| Well-defined task | A task where the facts, the format and the audience are all already clear before writing starts. |
| Checking cost | The time and attention needed to verify an output is correct before it is used. |
| Recurring task | A task repeated on a regular schedule, such as a weekly restock note, rather than done once. |
| Judgement-heavy task | A task where deciding what to say matters more than how the sentences are worded. |
| AI tool | A chatbot-style program that produces text, drafts or summaries in reply to written instructions. |
An AI tool saves time on tasks where turning known facts into organised text is the slow part, and checking the result against those same known facts is quick. It saves little or no time, and can even cost time, on a task where deciding what to say is the hard part, because instructing the tool clearly, then checking a judgement-heavy answer carefully, can take longer than simply thinking it through and writing it directly.
| What can differ | The shapes you will meet | Why it matters to you |
|---|---|---|
| Order type | An export shipment note for Order A has fixed, known facts; a domestic stock-program note may depend on a forecast that is still being finalised. | A task is well-defined when its facts are settled, whatever the order type. |
| Business model | A Make to Order (MTO) task is usually one-off, so the time spent instructing an AI tool is spent once; a Make to Stock (MTS) task, like a weekly restock note, is recurring, so a good instruction pattern pays off many times over; a hybrid order may need the confirmed and forecast parts drafted separately. | A recurring task rewards the time spent getting the instruction right the first time, more than a one-off task does. |
| How much judgement the task needs | Restating known facts in a clean format is low-judgement; deciding how to phrase a sensitive message to a buyer about a delay is high-judgement. | High-judgement tasks are usually faster and safer to write directly, then use an AI tool only to tidy the wording afterwards if needed. |
| How costly a checking mistake would be | An internal production note is cheap to correct if wrong; a note reaching Northline Apparel or a mill outside Delta Garments is costly to correct if wrong. | A high-cost task deserves careful checking regardless of who wrote the first draft, but a low checking cost is what makes asking AI for a first draft actually save time. |
| Product family | The specific facts differ (a towel order's GSM and pile height, a bedding order's fabric approval), but the same three questions decide the starting point for any product family Delta Garments makes. | The decision method does not need to be relearned for each product. |
Karim has two writing tasks. The first is a short internal note restating Order A's known facts for the next shift: 15,000 pieces, fabric 180 GSM, agreed FOB USD 3.20 per piece, current stage sewing. The facts are already settled and written down; the note just needs to be organised clearly. Karim asks an AI tool for a first draft, checks the four facts in under a minute, and sends it.
The second task is a message to Cedar & Co., a domestic Make to Stock buyer, explaining why this week's restock quantity is being reduced because of a fabric delay. The facts exist, but the hard part is judgement: how much detail to give, how to keep the relationship steady, and what commitment to make about next week. Karim writes this one himself. He could ask an AI tool to draft it, but checking whether the tone and the implied commitment are right would take longer than writing the three sentences directly, because the value here is in the judgement, not the wording.
The worked numberKarim tracks his own time on both tasks for one afternoon.
Asking an AI tool for a first draft did not save time on every task; it saved time on the one task where the facts were already settled and the check was quick, and it would have cost more time on the task where the value was in Karim's own judgement about tone and commitment.
Once asking an AI tool for a first draft becomes a habit, it is tempting to use it on every task, including a sensitive message where the real work is deciding what to say, not how to phrase it. A common mistake is spending time instructing and then carefully checking a judgement-heavy AI draft, when writing three honest sentences directly would have taken less time and carried the judgement clearly, because it came from the person actually accountable for it.
An AI tool is fastest to use well on a task where the facts are already settled and checking the result takes only a minute. It is not automatically faster on a task where deciding what to say is the hard part, because instructing the tool and checking a careful answer can take longer than writing it directly. Before starting a task, ask whether the facts are settled, whether checking will be quick, and whether the value is mostly in the words or the judgement, then choose the faster real path, not the habitual one.
Karim's assistant noticed that asking an AI tool for a first draft worked well on the Order A production note and suggests using the exact same approach on every writing task this week, including the Cedar & Co. restock message about the fabric delay.
Should Karim apply the same AI-first-draft pattern to the restock message? Decide before opening the answer.
No. The two tasks differ on the three deciding questions, so the same starting point does not fit both.
1. The production note's facts were already settled (15,000 pieces, 180 GSM, USD 3.20 FOB, current stage); the restock message depends on a judgement about tone and commitment that is not a fixed fact to restate. 2. Checking the production note took under a minute, because it only required matching four known facts; checking a judgement-heavy AI draft for tone and implied commitment would likely take longer than writing the message directly. 3. The value in the restock message is in Karim's own judgement about how much to say and what to promise, not in how the sentences are organised. 4. Applying one successful pattern everywhere ignores the task's own shape. The decision needs to be made task by task, using the same three questions each time, not by habit.
The key point is that a method that saved time on one task does not automatically save time on a different task with a different shape.
Order A's total FOB value stays the fixed reference figure for this course:
The restock message itself carries no large single figure like Order A's, but the cost pattern is the same one that protects an order worth USD 48,000 elsewhere in this course: time spent on the wrong tool for the task is time not spent on the judgement that actually matters. If Karim had spent seven minutes instructing and checking an AI draft for the restock message, instead of four minutes writing it directly, he would have spent three minutes longer to get a result that still needed his own judgement layered on top, on a task where that judgement was the entire point.
Use the T18-C02 Day-to-day AI-use checklist workbook. For each task this week, record whether you asked an AI tool for a first draft or wrote it yourself, and note briefly why. Add Karim's two tasks as example rows, then track your own choices for one full week.
[Download the Day-to-day AI-use checklist workbook](../../templates/T18-C02-day-to-day-ai-use-checklist.xlsx)
1. Why did Karim write the Cedar & Co. restock message himself instead of asking an AI tool for a first draft?
2. A merchandiser at a hybrid-order factory has a recurring weekly task: restating a confirmed portion and a forecast portion of the same order in a short internal note. Which starting point best fits the three deciding questions?
| EN | AR | FR | TR | Meaning |
|---|---|---|---|---|
| First draft | مسودة أولى | Premier brouillon | İlk taslak | The first attempt at a piece of writing, produced either by a person or by an AI tool, meant to be checked and improved. |
| Well-defined task | مهمة محددة بوضوح | Tâche bien définie | İyi tanımlanmış görev | A task where the facts, the format and the audience are all already clear before writing starts. |
| Checking cost | تكلفة التحقق | Coût de vérification | Kontrol maliyeti | The time and attention needed to verify an output is correct before it is used. |
| Recurring task | مهمة متكررة | Tâche récurrente | Tekrarlayan görev | A task repeated on a regular schedule, such as a weekly restock note, rather than done once. |
| Judgement-heavy task | مهمة تعتمد على الحكم المهني | Tâche à fort enjeu de jugement | Değerlendirme ağırlıklı görev | A task where deciding what to say matters more than how the sentences are worded. |
| AI tool | أداة ذكاء اصطناعي | Outil d'IA | Yapay zekâ aracı | A chatbot-style program that produces text, drafts or summaries in reply to written instructions. |
Next you will learn how to keep a human check in place on any AI-assisted output that leaves your own team, seen through a mill's own hand-off of material specifications. Back to the course map