The night before, open your AI tool, tell it what the block is for, and have it give you one thing: the first move you'll make when the block starts. That's it. The calendar block is the container. AI's job is to make sure you know exactly what goes in it, so you don't waste the first twenty minutes deciding.
Most deep work advice tells you to protect the time. Fewer people talk about what happens when the time arrives and you sit there, staring, not sure where to begin. That gap is where good intentions go to die. Here's how to close it.
Why the calendar block isn't enough on its own
A blocked calendar is a promise about when. It says nothing about what or how. So you show up at 9:00 for "Deep Work: Q3 proposal," and your brain fires off three questions: What part? Starting from what? Do I have everything I need? That's friction, and friction at the start of a session is the most expensive kind. It's the moment you're most likely to open a browser tab and vanish for half an hour.
AI helps here, but not the way people usually reach for it. You're not asking it to do the work. You're asking it to clear out the decisions sitting between you and the work.
The five-minute prep routine
Do this the night before, or first thing that morning before the block starts. Five minutes, not fifty.
Step 1: Name the real outcome
Look at the block and write one sentence: what does "done" look like for this session? Not the whole project. Just this block. "A rough draft of section two" beats "work on the proposal." Give that sentence to your AI as the goal. If you can't state the outcome in a sentence, that's your signal the block is too vague, and sorting it now saves you the pain later.
Step 2: Dump the context
Give the AI what it needs to be specific: the deadline, who it's for, what stage you're at, any constraints. This is where generic prompts fall apart. Vague input gets you a lecture. Specific input gets you something you can use.
Step 3: Ask for the first move, not the plan
Here's the part people skip. Don't ask for a full plan. Ask for the single first action you'll take when the timer starts. One concrete thing. A first move is momentum. A full plan is just another document to read and second-guess.
Step 4: Pre-load the obstacles
Ask the AI what's likely to stall you mid-session and what to do when it hits. Missing a number? Decide now to drop a placeholder and keep moving. Tempted to research? Decide now to leave a parking-lot note instead of opening ten tabs. Naming the stall in advance is what stops it from ending your session.
Step 5: Write the two-line brief
Collapse everything into two lines you'll see when the block opens: the outcome, and the first move. Paste them into the calendar event's notes field. When 9:00 arrives, you don't think. You read two lines and start.
Where a prompt pack saves you the setup
The routine works. The catch is that writing good prep prompts from scratch, every single day, is its own small tax, and after a week most people quietly stop. That's exactly why I built the Focus Defender Deep Work Prompt Pack. It's a Notion template with ready-made prompts for naming the outcome, pulling the first move, and pre-loading the stalls, so the five-minute routine becomes fill-in-the-blank instead of blank-page. If you'd rather build your own, the steps here are the whole method. If you'd rather just run it, that's what the pack is for.
Fitting this into a calendar-first day
The whole point of a calendar-first day is that decisions get made once, in advance, so your working hours stay decision-free. This prep routine is the same idea, one layer down. Your calendar decides when you work. Your five-minute prep decides what the first move is. By the time you're in the block, both answers are already there, and all that's left is the work.
Run it on your one or two most important blocks, not every event on the calendar. Prep is only worth it for the sessions where a slow start actually costs you something.
FAQ
How far in advance should I prep a deep work block?
The night before is ideal, because your morning brain gets to start instead of plan. If that's not realistic, do it in the ten minutes before the block. Same-day is fine. Skipping it entirely is what hurts.
Won't asking AI to plan my session just create more busywork?
It will if you ask for a full plan. Keep it to one outcome and one first move. The goal is fewer decisions, not a longer document. If your prep is more than two lines, you've over-planned.
What if the block gets interrupted or cut short?
Because you defined a first move instead of a rigid plan, you can start right away and get real work done in whatever time you actually get. A ten-minute head start on the right thing beats an hour of figuring out where to begin.
Do I need a special tool, or will any AI work?
Any capable AI chat tool works. The value is in how you prompt it, not which one you use. A prompt pack helps mainly by removing the daily setup cost, so the habit survives past the first week.