One Thing at a Time

---

Intro

It feels efficient to ask for several things at once. In practice, it usually produces a response that does all of them slightly badly. One thing at a time is faster in the end.

---

What is happening

When you put three tasks in one message, the AI tries to serve all three simultaneously. It has to balance how much space to give each one, which order to tackle them in, and how to keep them from bleeding into each other. The result is often a response that feels like a first rough draft of all three — none of them quite what you wanted.

When you send each task as its own message, the AI gives it its full attention. You read the response before moving on. If something is off, you can fix it before it shapes the next step. You stay in control of the direction.

This is not about the AI being limited. It is about how any focused work goes better when you do one thing at a time.

---

The exercise

Pick a task that has three parts. It can be anything from your actual work. If nothing comes to mind, use the example below.

Example task:
You want to use an article you have read to: summarise it, write a social media post about it, and draft a short email to your team with the key points.

Step 1: Send everything at once

Copy this message and personalise the brackets before sending:

Can you summarise this [article / document / piece of text], then write a social media post about it, and also draft a short email to my team with the key takeaways? The text is: [paste your text here]

Send it. Read the whole response. Notice how each part feels.

Step 2: Start a new conversation and send them separately

Open a new conversation so you are starting fresh. Then send these three messages one at a time, reading and checking each response before sending the next.

Message 1:

Summarise this [article / document / piece of text] in 5 bullet points. Keep each bullet under 20 words. [paste your text here]

Read the summary. Is it accurate? Does it cover the right things? If it missed something, ask for a change before moving on.

Message 2 (after you are happy with the summary):

Using that summary, write a social media post. Keep it under 280 characters. Tone: [choose one — casual / professional / curious].

Read it. Adjust if needed.

Message 3 (after you are happy with the post):

Now draft a short email to my team sharing the 3 key takeaways. Keep it under 100 words. Start with the most important takeaway.

Read it.

Step 3: Compare

Look at the results from Step 1 and Step 2. Which version of each piece is more useful? Where did the single-message version cut corners?

---

COPY — PERSONALISE — USE

Here is a single-purpose message template you can use for any task going forward:

I need to [verb: summarise / draft / list / explain] [the specific thing]. [One relevant detail about your situation]. [Stopping condition: e.g. "Keep it under 100 words" or "Give me 3 options"].

How to personalise it:

[verb] — pick the action that fits: "summarise" (shorten it), "draft" (write a first version), "list" (give me options), "explain" (help me understand).

[the specific thing] — be specific. Not "my email" but "my reply to the client complaint I received yesterday."

[one relevant detail] — one fact that changes how the AI should approach this. "The audience is non-technical." "My manager prefers plain language." "This is for internal use only."

[stopping condition] — how you will know it is done. "Under 100 words." "3 bullet points." "One paragraph."

---

What good looks like

Each of your three separate responses should feel more focused and complete than its equivalent section in the combined response. The effort of sending three messages instead of one should be offset by spending less time editing the results.

---

If this did not work

If the separate messages did not produce noticeably better results, check whether each message was truly single-purpose. "Summarise this and also tell me what I should do next" is still two tasks. Keep each message to exactly one verb and one target.

If the responses are still coming back longer or vaguer than you want, B-06 covers how to add a stopping condition that keeps the AI within bounds.

---

Next

Now that you are sending one thing at a time, B-06 shows you how to tell the AI exactly when to stop — so you get the size and shape of response you actually want.