AI tools save time, but constant prompting, waiting and checking can quietly wreck your focus. Learn how attention residue works and try a simple Pomodoro routine for working with AI.
🤖 The New Distraction Nobody Planned For
A few years ago, the biggest enemies of focus were email and social media. In 2026 there's a new one, and it sits right inside our work: the AI chat window.
The pattern goes like this. You write a prompt. The answer takes 30 seconds, or an agent runs for a few minutes. That gap feels too short for real work, so you check your messages. Then a news tab. Then the answer is ready, you skim it, ask a follow-up and wait again.
After an hour you've been "working with AI" the whole time. And yet you feel tired, scattered, and not sure what you actually finished.
You're not imagining it. There's a name for what's going on.
🧩 What Is Attention Residue?
In 2009, researcher Sophie Leroy described attention residue. When you switch from Task A to Task B, part of your attention stays stuck on Task A. Your mind keeps chewing on it in the background, and you do worse on Task B. The effect is stronger when Task A is left unfinished.
Now look at a typical AI workflow. Every wait is a small unfinished task. Every "let me check something while it thinks" is a switch. Soon you're carrying residue from three or four things at once: the prompt, the other chat, the tab you opened, the reply you only half read.
AI didn't invent task switching. It just made it faster, more frequent and easier to justify, because all of it looks like work.
🔍 Signs AI Is Breaking Your Focus
- You open a new tab every time the model is "thinking".
- You re-prompt five times instead of spending two minutes deciding what you actually want.
- You have six chat threads open and can't remember which one had the good answer.
- You accept an answer without really reading it, because your mind has already moved on.
- At the end of the day you've produced a lot of text but finished very few tasks.
If three or more of these sound familiar, the fix isn't to stop using AI. It's to give it a structure.
🍅 The AI Pomodoro: A Simple Routine
This routine fits inside one Pomodoro. It splits AI work into clear phases, so your brain isn't trying to do everything at once.
1. Think first, without AI (2–5 minutes)
Before you open any chat, write down in plain words: what's the goal, what does "done" look like, and what do you already know. Two or three lines are enough. This is the part AI can't do for you, and it makes your prompts much better.
2. Batch your prompts
Instead of question, wait, question, wait, write one complete prompt with all the context, or send several requests at once. Fewer round trips means fewer switches.
3. Decide what you'll do while it works
The wait is the dangerous moment. Pick a waiting task that belongs to the same project:
- Sketch the next step on paper
- Re-read the brief or your own notes
- Write the part of the work you don't want AI to touch
- Or simply look away from the screen and rest your eyes
What you don't do: open email, chat, news or social media. Those are different tasks, and they leave residue.
4. Review with full attention
When the answer arrives, read it properly. Check the facts, run the code, compare it with the goal you wrote in step 1. AI output still needs a human who is paying attention. That's you.
5. Close the loop before the break
Before the timer rings, write one line: "Next: …". Later research by Leroy and Theresa Glomb suggests that a quick "ready-to-resume" plan like this helps reduce attention residue when you're interrupted. It also makes the next session much easier to start.
🗂️ Keep "AI Time" and "Thinking Time" Separate
Not every task needs AI. A rough split that helps:
- Good for AI sessions: first drafts, summaries, boilerplate code, reformatting, brainstorming lists, explaining an unfamiliar topic.
- Better as AI-free deep work: deciding what matters, writing your key argument, designing a solution, hard debugging where you need to truly understand the problem, anything personal.
One way to plan your day: start with one or two AI-free Pomodoros for the thinking-heavy part, while your head is fresh. Then switch to AI-assisted sessions for drafting and execution.
Mini-scenario: you need to write a project proposal. Session 1, no AI: list the three points the proposal must make and who will read it. Session 2, with AI: generate a first draft from those points and ask for weak spots. Session 3, no AI: rewrite the draft in your own voice. Three focused rounds instead of an afternoon of chatting back and forth.
⚙️ Setting It Up in Pomotastic
- Put the session goal in the task field, for example "Draft section 2 with AI and review it". It's your anchor every time you come back from a wait.
- Match the interval to the work. Prompt-heavy, fast tasks often fit 15–20 minutes. Work with lots of reading and review may need 30–40. You can change it in Settings.
- Turn on notifications, so the end of a session reaches you even when you're deep in another window.
- Keep breaks away from the screen. You've just spent 25 minutes reading generated text. Stand up, stretch, look at something far away.
That last point matters to me personally. I built Pomotastic after long hours at the laptop gave me neck and back pain. AI tools can make those hours even longer, because there's always one more prompt to try. A timer is a good reminder that you're a person sitting in a chair, and that person needs to move.
📋 Quick Checklist
- ✅ Goal written down before the first prompt
- ✅ One chat thread per task
- ✅ A waiting task chosen in advance
- ✅ No email or social feeds while the model works
- ✅ Every answer reviewed, not skimmed
- ✅ "Next: …" written before the break
- ✅ Breaks taken away from the screen
💬 Final Thoughts
AI can make you faster. But speed without focus just means more half-finished work. Give AI a clear place inside your Pomodoro, protect the waiting moments, and keep the most important thinking for yourself.
Start a timer, write down your goal, and try one AI Pomodoro today. 🚀
❓ FAQ
What is attention residue?
It's the part of your attention that stays stuck on a previous task after you switch to a new one. The term comes from Sophie Leroy's 2009 research. It lowers your performance on the new task, especially when the old one was left unfinished.
Does using AI make you less focused?
Not by itself. The problem is the work pattern around it: frequent short waits and constant switching between chats, tabs and apps. A routine with batched prompts and a planned waiting task keeps the benefits without the scatter.
What should I do while waiting for an AI response?
Stay inside the same task. Plan the next step, re-read your notes or rest your eyes. Avoid email and social feeds, because they pull your attention to a completely different task.
How long should a Pomodoro be for AI work?
Start with the classic 25 minutes. If you spend a lot of time reviewing output, 30–40 minutes may suit you better. For short, fast prompt work, try 15–20 minutes.