5 AI Prompts for Single-Tasking (Copy and Use Today)

The problem with multitasking is not that people lack willpower — it is that the brain treats an unfinished task as an open loop, and open loops compete for attention even when you are not consciously thinking about them. Research by Masicampo and Baumeister (2011) showed that simply having a concrete plan for an unfinished task suppresses intrusive thoughts about it. These prompts exploit that mechanism: by offloading the queue to an AI at the start of a session, you close the loops that would otherwise leak into your focus block.

These prompts implement the One Thing Lock protocol. Copy them, fill in the brackets, and use them as-is. Adjust language to your own style once you have run them a few times.

Prompt 1: Session Setup (The Lock)

Use when: You are about to start any focus block and have competing tasks or open items in your head.

I'm starting a 45-minute One Thing Lock.

My task for this block: [specific output, one sentence]

After this block I need to handle:
- [item 1]
- [item 2]
- [item 3]
(add as many as needed)

Please:
1. Confirm my task back to me in one sentence
2. Hold the queue and give me a briefing when I return
3. If I ask you about anything unrelated to [task name] during this block, remind me to stay on task

I'll check back at [time].

The confirmation in step 1 is not redundant. Having the AI reflect the task back activates the goal intention in a way that writing it privately does not.

What good output looks like: The AI restates your task in a single, clear sentence (not a paraphrase that expands the scope) and acknowledges the queue items without editorializing. If the AI starts suggesting alternative tasks or asking what you want to do first, the prompt failed — it should be confirming the lock, not opening a planning conversation.

Prompt 2: Mid-Block Task Support

Use when: You need AI help during the session on a specific, task-relevant problem — not to explore adjacent ideas.

I'm mid-block on [task name].
Specific help needed: [precise question or request]
Keep the response focused—I'll be back on task immediately after.

The framing keeps responses tight and discourages open-ended exploration. During a lock-in, you want answers, not conversations.

What good output looks like: A direct answer to your specific question, short enough to read in under thirty seconds. A sign the prompt failed: the AI responds with “Great question — there are several ways to think about this…” and delivers three paragraphs. That is a conversation starter, not a task-focused answer. If that happens, re-run with your question made more specific and add “one paragraph only.”

Prompt 3: Mid-Block Interruption Capture

Use when: Something urgent arrives — a message, a stray thought, a request from someone — that you need to note without actually switching tasks.

Interruption capture (do not break my focus block):
Item: [what arrived]
Category: [urgent / queue for later / discard]
Add to my queue if relevant. I'll assess at unlock.

If you are genuinely unsure whether the item is urgent, default to “queue for later.” Most things that feel urgent in the moment are tolerant of a forty-five minute delay.

What good output looks like: A brief acknowledgment (“Added to your queue. Back to your block.”) with no further prompting. A sign the prompt failed: the AI responds to the content of the interruption — starts answering the question from the email, or weighs in on whether you should handle it now. If that happens, the prompt needs the phrase “do not engage with the content of this item” added explicitly.

Prompt 4: Task Definition Check

Use when: You have a block scheduled but your task statement feels fuzzy — you know what area you want to work in but not what “done” looks like.

I'm planning a 45-minute focus block. My current task definition is:
"[your current task statement]"

Is this output-defined and completable in 45 minutes? 
If not, suggest a more specific version.

Vague task definitions are the most common cause of unproductive focus sessions. This prompt catches them before they waste the block.

What good output looks like: The AI either confirms the task is specific enough and explains why, or proposes a tighter version — something like “Write the outline for section 2 of the proposal (three main points, one paragraph each)” rather than “Work on the proposal.” A sign the prompt failed: the AI confirms a vague task is fine, or offers multiple alternative task definitions without recommending one. If you get four options, push back: “Pick the one that would produce the most useful output in 45 minutes.”

Prompt 5: The Unlock Review

Use when: Your focus block ends and you need to transition deliberately rather than drift into whatever feels easiest next.

My 45-minute block is done.

Task status: [complete / here's what I got through and what remains]
New captures during block: [list anything from your notepad]

Please:
1. Give me the queue briefing from setup
2. Add my new captures to the queue
3. Tell me the top 1-2 items to handle next
4. Ask me if I want to start another lock or take a break

The final question is important. It creates an explicit decision point rather than letting momentum pull you automatically into the next activity.

What good output looks like: A clean queue briefing (the items you logged at setup, in order), your new captures integrated, and a clear recommendation on what to do next. A sign the prompt failed: the AI loses the queue from setup — this usually means the conversation context was too long or you started a new chat. Solution: keep Prompt 5 in the same conversation thread as Prompt 1, or paste the queue items again in Prompt 5’s context field.

Common Pitfalls

Pitfall: Using vague task names in Prompt 1. “Work on the report” is not a lockable task — there is no moment when you can confirm it is done. Fix: before running Prompt 1, ask yourself what one specific output would exist if the session went perfectly. Use that as the task definition.

Pitfall: Starting a new chat between Prompt 1 and Prompt 5. The queue handoff only works if both prompts run in the same conversation thread. If you open a new chat for mid-block help, the unlock review will not have context. Fix: use Prompt 2 and 3 as follow-ups in the Prompt 1 thread, and open a separate fresh chat if you need unrelated AI help during the block.

Pitfall: Running Prompt 4 as a substitute for Prompt 1. Task definition check and session setup are two different operations. Prompt 4 sharpens the task; Prompt 1 locks it and transfers the queue. Skipping Prompt 1 after running Prompt 4 leaves the open-loop problem unsolved — you have a better task definition but no offloaded queue.

Pitfall: Treating the AI’s “next steps” recommendation in Prompt 5 as binding. The unlock review tells you what the data suggests, not what you should do. If the AI recommends handling queue item 2 next but you know item 1 has a hard deadline, override it. The value of the review is the briefing, not deference to the ranking.

How to Build These Into a Habit

The most effective approach is to keep Prompts 1 and 5 in a note or document you can open instantly. The two to three minutes of friction involved in writing them from scratch each session is enough to cause skipping.

Once you have run both prompts five or six times, you will likely start customizing them naturally — shortening the language, adjusting the queue structure, adding fields relevant to your work context. That adaptation is a sign the protocol is becoming habitual.

Copy Prompt 1 into a note right now, fill in today’s most important task and your current mental queue, and run your first session before the end of the day. For the research and framework behind the One Thing Lock — including why 45 minutes is the default block length and how to handle recurring interruption patterns — see the Complete Guide to Single-Tasking with AI Support.

Related:

• How to Single-Task with AI Support
• The Complete Guide to Single-Tasking with AI Support
• The Single-Tasking Framework with AI
• Beyond Time Single-Tasking Walkthrough

Tags: AI prompts, single-tasking, focus sessions, One Thing Lock, quick win