My Morning Routine With AI: How I Save 2 Hours Every Day

I am not a 5 AM productivity guru. I don't meditate for 40 minutes or journal three pages or take cold showers. My morning routine is 35 minutes long and built around three AI tools doing the parts of morning prep I'm bad at.

The result has been the most consistent productivity gain I've had in years. I save roughly 2 hours of work per day — not by working faster, but by deleting the decision fatigue and admin scrambling that used to fill my mornings. Below is exactly what I do, in order, with the prompts I actually use.

This isn't a "look how disciplined I am" post. It's a "here's a routine that's so easy it's hard to skip" post.

The core idea

Most morning routines fail because they're too long, too virtuous, or both. People try to stack meditation, exercise, journaling, and reading into the first 90 minutes of the day, then quit when they oversleep on day six.

My approach is the opposite: do the smallest set of things that meaningfully reduces friction for the rest of the day. AI handles the parts that used to drain me — triage, planning, deciding what matters — leaving the human work (actually doing things) easier.

The whole routine takes 35 minutes. Some days less.

The 35-minute routine

Time	Activity	Tool
0–5 min	Coffee + 24-hour world brief	Gemini (free)
5–15 min	Inbox triage with AI assist	Superhuman + Claude
15–25 min	Daily plan from yesterday's notes	Claude + Notion
25–30 min	Calendar review + reschedule	Reclaim + Google Calendar
30–35 min	First-task warmup	Claude or no tool

Now the actual mechanics for each.

Phase 1: World brief (5 minutes)

The first thing I do — before I open email — is open Gemini with this prompt:

Give me a 5-bullet briefing on what happened in the last 24 hours that would be relevant to a [your role/industry] in [your location]. Focus on: tech, business, AI, and anything that materially affects [my industry]. Skip celebrity news, sports, weather. Cite sources.

This replaces the 25 minutes I used to spend doomscrolling Twitter "just to catch up." Gemini's web access means it pulls live news. The cited sources let me click through if a bullet matters; otherwise the bullet itself is enough context.

The discipline here is closing the tab after reading the briefing. No clicking through to argue with strangers. Five minutes, done.

Why this saves time later: I used to lose chunks of the morning to news rabbit holes. The AI brief gives me the awareness signal I was actually after, in a fraction of the time, with zero algorithm-bait.

Phase 2: Inbox triage (10 minutes)

I open Superhuman and run my standard triage routine. Anything that takes under 60 seconds gets handled immediately. Anything that takes longer goes into one of three buckets:

• Reply today (flagged with star)
• Reply this week (snoozed to a specific day)
• Read for context (split-archived after a quick skim)

For longer replies that I'll handle later, I draft a 30-second outline of what I want to say in Claude:

I need to reply to this email later today. Help me draft a one-sentence reply outline I can come back to: (1) what I want to say at the top, (2) what I want to leave out, (3) the call-to-action or close. The email: [paste]

Claude returns a tiny outline. I save it in the email's Notes field in Superhuman. When I sit down to write the reply later, I'm not starting from a blank page — I'm finishing a sketch.

Why this saves time later: Drafting a reply from scratch takes 8–15 minutes. Drafting from a 3-line outline takes 3–5. Across 10 emails a day, that's an hour saved.

Phase 3: Daily plan from yesterday's notes (10 minutes)

This is the most valuable part of the routine.

I keep a "Daily Notes" page in Notion. Every working day has its own page, with whatever scratch thoughts, tasks, decisions, and meeting notes I dumped during the day before.

In the morning, I paste yesterday's page into Claude with this prompt:

Below are my notes from yesterday. Extract: 1. Anything I committed to doing (with deadlines if mentioned) 2. Decisions I made that need follow-up actions 3. Open questions I left dangling 4. Anything that suggests today's priorities

Then propose a 3-item priority list for today, ordered by importance, with my reasoning.

Notes: [paste]

Claude returns a structured pickup list. I read it, edit it (sometimes I disagree with the priority order), and paste the final 3-item list into today's Daily Notes page.

That's it. Three priorities. Not ten. Not twenty. Three.

Why this saves time later: The "what should I work on first" decision used to consume real morning energy. Now it's made for me with full context, in 90 seconds. The 3-item limit is non-negotiable — anything more is a wishlist, not a plan.

Phase 4: Calendar review + reschedule (5 minutes)

I open Google Calendar (with Reclaim integration) and do three things:

1. Look at every meeting today. Confirm they all need to happen. Cancel or reschedule any that don't.
2. Block protected time for each of the 3 priorities from Phase 3. If the priorities don't fit on the calendar, I have to drop one — there's no point committing to work I won't have time for.
3. Glance at tomorrow to see if anything needs prep today.

Reclaim handles the "block time around meetings" part automatically based on rules I set up. I just review and adjust.

Why this saves time later: Most days where I felt unproductive turned out to be days where my calendar was full of inherited meetings and zero protected time for actual work. Building protected time into the calendar in the morning prevents the late-afternoon panic of "I haven't done anything that mattered today."

Phase 5: First-task warmup (5 minutes)

This is the only phase that doesn't always involve AI.

For whichever priority is the first one on my list, I do one small concrete thing to get started. Open the document. Read the first paragraph of yesterday's draft. Write a single sentence. Open the codebase to the file I need to edit.

If the task is genuinely intimidating (a hard email, a complex draft, a decision I've been avoiding), I'll sometimes use Claude to lower the activation energy:

I need to start on [task]. The hard part is [the specific reason it feels hard]. Help me identify the first 10-minute step that would meaningfully reduce my dread about this.

Claude is surprisingly good at this. It usually proposes something embarrassingly small — "open the document and outline three bullet points" — that I can absolutely do, which breaks the block.

Why this saves time later: The first 30 minutes of any deep-work session are the hardest. Doing 5 minutes of warmup work in the morning means when I sit down for real later, I'm not starting from cold.

What I deliberately don't do

A few things missing from this routine that you might expect:

No social media check. Nothing about my morning is improved by Twitter, LinkedIn, or Instagram before 9 AM. Notifications on my phone for these are off until 5 PM.

No reading. I love reading but it's an evening activity. Mornings are for clearing the deck.

No exercise. I exercise around lunch. Trying to stack a workout into the morning routine made me skip both.

No journaling. I tried it. It made me dread mornings. I journal occasionally on weekends, which is enough.

No lengthy planning system. Three priorities and a calendar review. That's the whole plan. Anything more elaborate becomes its own form of procrastination.

The point of this routine isn't to do more. It's to clear the runway so the actual work of the day starts faster.

What changed for me when I adopted this

Three things shifted in the first month:

1. I stopped feeling "behind" by 11 AM. The 3-priority list is shorter than my old to-do list, but I actually finish it. Finishing creates momentum; not finishing creates dread.

2. My evenings got better. The mental load of "what didn't I get to today" dropped because the daily plan was honest about capacity from the start.

3. I noticed my phone less. The world brief replaced the doomscroll. The triaged inbox replaced the constant "is there a fire?" check. Both of those, repeated daily, retrained my reflex to grab my phone for stimulation.

How to start your own version

Three pieces of advice if you want to adapt this:

1. Start with one phase. The world brief is the easiest entry point — it takes 5 minutes and produces an immediate sense of agency. Add the inbox phase a week later. Add the planning phase the week after that.

2. Customize the prompts to your context. The Claude planning prompt above works because Claude knows my style after months of use. Yours will need to reference your industry, role, and the kind of decisions you make.

3. Track time once, then stop. For one week, time how long each phase takes. After that, don't time it. The metric matters once to confirm the routine isn't bloated; tracking forever turns it into homework.

What this is not

This isn't a routine that will turn you into a productivity machine. It won't make you more disciplined or focused or rested. The benefits compound — modest savings each morning add up over weeks — but day one will not feel revolutionary.

What it does is reduce the small frictions that accumulate into "I had a busy day but got nothing done." Over months, that's the most valuable change a morning routine can produce.

The takeaway

Stop trying to build a morning routine that requires willpower. Build one that requires 35 minutes and three AI tools.

The best routine is the one you'll do tomorrow, and the day after, and three months from now. Boring beats virtuous. Repeatable beats impressive. AI handles the parts you'd procrastinate on, so the parts you actually have to do feel smaller when you get to them.

That's the whole secret. Two hours saved daily, from 35 minutes of consistent setup. The math works because you're not buying speed — you're buying clarity.