All systems
System 04AI Productivity Systems

A calm weekly rhythm — without another productivity app.

Use AI to triage your inbox and calendar, surface the few things that actually matter, and ship a one-page plan you can keep open all week.

~2h saved / week
Beginner
ChatGPT · Notion · Apple Notes
What this system solves

By Wednesday the plan you wrote on Sunday is already wrong.

It's rarely a discipline problem. It's that the inputs — calendar, inbox, project notes, half-finished commitments — never get processed in one place, on a schedule.

  • Information overload
    Five tools, three notebooks, an inbox. The plan dies in the gap between them.
  • Decision fatigue
    Every Monday morning you're choosing what matters from scratch, with no system.
  • Calendar drift
    Meetings expand to fill the week. Focus blocks are the first thing to get cut.
  • Vague priorities
    'This week is about the launch' — but no concrete commitments behind it.
  • No weekly close
    Friday ends without a clean handoff to next week, so Sunday starts heavy.
Who it's for

Built around real workflows.

Founders and operatorsSenior individual contributorsConsultants juggling multiple clientsManagers running their own week + a team'sAnyone tired of productivity-app onboarding

If your week is shaped by five contexts and the plan you wrote on Sunday rarely survives Tuesday — this system is for you.

Workflow overview

Five steps. About twenty minutes on Sunday.

The whole point is that it's repeatable. Run it the same way every week and the compounding effect is the entire return.

01
Pull the inputs

Get everything you've already written down into one place.

How to run it: Open your calendar, inbox, and project notes for the week. Paste the relevant text — meetings, threads, commitments — into a single doc.

02
Surface the priorities

Let the model triage the week down to what matters.

How to run it: Prompt the model with the inputs and ask for 3 outcomes that would make the week a success, plus 5 specific commitments behind them.

03
Review the calendar

Make sure the week's shape matches the priorities, not the other way round.

How to run it: Ask the model to flag meetings that don't map to the week's outcomes, suggest which to decline, defer, or shorten.

04
Block focus time

Protect 2–4 hours of deep work before the meetings claim them.

How to run it: Have the model propose 2–3 focus blocks based on the priorities and existing calendar. Drop them straight onto the week.

05
Ship a one-page plan

Leave Sunday with a single page you can keep open all week.

How to run it: Generate a clean weekly plan — outcomes, commitments, focus blocks, and what you're explicitly not doing this week.

06
Close the week (Friday)

End with a clean handoff to next Sunday's planning session.

How to run it: Paste what got done into the model and generate a 5-line review: what shipped, what slipped, what to carry over.

Example prompts

The prompts that run the system.

Tested across GPT-5, Claude Sonnet, and Gemini 2.5. Replace anything in {curly braces} before sending.

Sunday triage
Here are my inputs for the week of {WEEK}:

Calendar:
"""
{CALENDAR}
"""

Inbox & threads:
"""
{INBOX}
"""

Open commitments / project notes:
"""
{NOTES}
"""

Produce:
1. The 3 outcomes that would make this week a success.
2. The 5 specific commitments behind them (each tied to an outcome).
3. The things I should explicitly NOT do this week.

Be honest. If too much is on the list, say so and suggest what to defer.
Calendar review
Below is my calendar for the week and the 3 outcomes I want.

Outcomes:
{OUTCOMES}

Calendar:
"""
{CALENDAR}
"""

For each meeting:
- Map it to one of the outcomes, or mark it "unmapped"
- Suggest: keep, shorten, defer, or decline (with one-line reasoning)

Then propose 2–3 focus blocks (90+ min) for the unmapped time, anchored to the most important outcome.
Friday close
Here's what actually happened this week:

"""
{WEEK_SUMMARY}
"""

In 5 lines:
1. What shipped
2. What slipped, and why
3. What I learned about how I work
4. What carries over to next week
5. One thing I should stop doing

Tone: calm, honest, no self-flagellation.
Example output

What the system actually produces.

Week of Mon 18 May · One-page planWeekly plan
OUTCOMES
1. Ship the v2 onboarding flow to staging by Thursday.
2. Close the loop with the three Q2 candidates.
3. Land the board update — one page, sent Friday AM.

COMMITMENTS
- Mon: write onboarding spec → Notion (90 min focus block)
- Tue: design review with Mara, then build (am)
- Wed: candidate calls × 3 (pm)
- Thu: ship v2 to staging, internal walkthrough at 16:00
- Fri AM: write + send board update

FOCUS BLOCKS
- Mon 09:30–11:00 (spec writing)
- Wed 09:00–11:00 (board update draft)
- Fri 08:30–10:00 (board update polish)

NOT THIS WEEK
- The pricing page rewrite
- Inbox zero
- Anything from the "someday" list
Common mistakes

Where this system tends to break.

Planning every hour
The point is the 3 outcomes and 5 commitments — not a colour-coded calendar nobody can keep.
Skipping the 'not this week' list
Without it, the plan is just a wishlist. Naming what you're cutting is half the value.
Letting AI decide priorities
The model can structure your inputs. Choosing what matters this week is still your call.
Skipping Friday close
No close means Sunday starts heavy. A 5-line review is enough — don't make it a ritual.
The human layer

AI structures the week. You decide what it's for.

Use the system to handle triage, structure and admin. Keep the parts that need you in the room — choosing the outcomes that matter, having the hard conversation, and saying no to the work that doesn't make the cut.

AI is for
  • Triage and synthesis
  • Calendar review
  • Drafting the plan
You are for
  • Setting priorities
  • Saying no
  • Owning the outcome
Issue 001Shipping weekly

Weekly practical AI systems for modern work.

Useful workflows, operational systems, and practical AI implementation strategies for modern professionals. No hype, no trend-chasing — just systems that work.

Next dispatch · Monday 06:00 CET

No spam. Unsubscribe anytime.