Build Your Personal Assistant Operating System

6.2

Run the Morning From One Command

One command replaces opening twelve apps

You say 'run my morning' and get: your three-layer morning brief, your email with urgent items and drafted replies, today's three priority tasks matched to your energy, relationship alerts for people you are meeting today, a resurfaced reading highlight relevant to today's work, and an energy check-in prompt.

One command, one output, and the modules are invisible. You see a single document that assembles everything you need to start the day.

This is the graduation exercise. Writing a clear prompt that combines multiple modules into one routine is the most advanced in the book. If you can do this, you have developed the core ability: describing what you want from a complex system in plain language.

A stylized teaching image showing multiple module outputs composing into one unified morning document — One prompt combines every into a single morning document.

Morning Orchestrator

Promise	A single command or scheduled trigger that runs every active in sequence, produces one combined output, and surfaces only the items that need your attention.
Sources	All active briefs, , and each module's connected sources (calendar, email, tasks, contacts, journal, and others you have configured).
Output	A unified morning document combining outputs from every active : brief, email , task list, relationship updates, decision reviews due, and any alerts from specialized modules.
Saves to memory	Orchestration log: which modules ran, which succeeded, which failed, and how long each took. items that were not addressed yesterday.
Approval boundary	Runs modules according to their individual approval boundaries. The itself does not add permissions: if a is read-only, it stays read-only in the orchestrated run.
Failure behavior	If a fails, the continues with the remaining modules and reports the failure in the output. A failed module does not block the rest of the morning run.
Review criteria	The combined output should be scannable in under five minutes. failures should be clearly reported. items from yesterday should be visible.
Graduation trigger	When the orchestrated run consistently produces useful output without manual intervention, schedule it to run automatically and deliver the result before you wake up.
Privacy tier	Inherits the highest of any included . If any module is local-only, the should also run locally.

Scheduling turns a one-time routine into a daily habit

Different modules run at different times. The schedule is how you combine them across the day:

A sample daily and weekly schedule

Time	What runs	What you get
7:00 AM	Morning brief + email + task priorities + relationship alerts + reading highlight + energy prompt	The unified morning document
12:00 PM	Email batch + energy check-in	A mid-day email update and an energy rating for the afternoon
5:00 PM	Email batch + journal prompt	End-of-day email summary and a prompt to write a few sentences about the day
Friday 4:00 PM		Cross- reflection with done list, patterns, carry-forwards, and three questions
1st of month	Monthly patterns + finance review + career reflection	Slow-moving trends, spending summary, and quarterly growth check

You design this schedule through conversation, not configuration. 'I want my morning routine to run at 7, my mid-day check at noon, and my on Fridays at 4.' In ChatGPT, you set this up using scheduled tasks. In Claude Code, you use scheduled routines. The assistant handles the scheduling mechanics.

keep the morning running when one source fails

The combined routine needs to handle partial failures gracefully. If the Gmail connector is down, the morning brief should still run using calendar and task data, not stall waiting for email. If the relationship has no data yet, the brief should skip the relationship section rather than producing an error.

Fallback instructions are part of the prompt: 'If a 's source is unavailable, note what is missing and produce the remaining sections. Never skip the entire morning document because one source failed.' For example, when the Gmail connector is down, the assistant notes 'email unavailable' and still delivers your calendar brief and task list.

Every 's approval rules carry into the combined routine

Each has its own , and the combined routine inherits all of them. The morning brief never reschedules. The email never sends. The task module never moves client work without asking. The relationship module never contacts anyone.

The prompt makes this explicit: 'Combine the modules. Keep each 's . Collect all suggested actions into a single approval section at the end of the morning document. I will review and approve actions in one place instead of module by module.'

When you open the morning document, you review one consolidated approval list instead of checking six separate outputs. The list shows every suggested action: send the drafted email reply to Sarah about the timeline, move the 3 PM standup to tomorrow for a report-writing block, reconnect with David who has been quiet for 34 days. You approve, modify, or dismiss each one.

The makes the morning document a podcast

Add the from the Turn Any Into Audio chapter and the morning document becomes a three-to-five minute audio summary. The narration rewrite flows through the executive summary, the top , today's three tasks, and any urgent flags. The approval section stays in text (you need to read and decide on those, not hear them in the background).

When you press play on the audio brief over coffee, you hear your calendar, your top tasks, and any urgent flags. By the time you sit down, you know what matters. The approval items are waiting in text for your review.

A visual dashboard is optional for readers who want one screen

For readers who want a visual interface, the capstone project is a single-page web dashboard that displays the morning document: today's schedule, email summary, task priorities, relationship alerts, energy trend, recent journal themes, and the approval queue. You ask a coding agent (Claude Code or Codex) to build it from the .

This is optional. If you ask a coding agent to build a single-page dashboard that reads from your , the morning brief, task list, and approval queue all appear on one screen. The text-based morning document is the core product. The dashboard adds a visual layer for readers who want it.

How this module breaks

What goes wrong	How you notice	What to correct once	What rule to save
Approval fatigue: the orchestrated run surfaces twelve items needing review, and you start approving without reading.	You approve actions reflexively because the volume of approval requests overwhelms your attention. A bad approval slips through.	Reduce the approval surface: promote well-tested modules to low-risk autonomous for safe actions. Reserve approval for high-impact items only.	Modules with a track record of correct output can graduate to autonomous for low-risk actions. The highlights only high-impact items for approval, keeping the approval list under five items.
Rule accumulation across modules makes the orchestrated output unpredictable.	Different modules apply overlapping or contradictory rules, producing inconsistent results. The email classifies a sender as urgent while the task ignores tasks from the same sender.	Run a cross- rule audit during the monthly review. Identify rules that overlap across modules and resolve conflicts.	The monthly rule review includes a cross- audit: list all rules from all modules, flag overlaps and contradictions, and resolve them into consistent cross-module policies.
One slow delays the entire morning run.	The orchestrated output arrives late because one took ten minutes to process while others finished in seconds.	Set a timeout for each . If a module exceeds its timeout, the skips it, includes a timeout notice in the output, and runs the remaining modules.	Each in the orchestrated run has a timeout (default: two minutes). Modules exceeding the timeout are skipped with a notice. The overall run should complete within the total timeout budget.

Your turn

Write your run-my-morning prompt

You'll produce a prompt that runs your morning from one command, with fallback rules and a consolidated approval section.

Why this exercise matters

This prompt is the capstone . It combines every you have built into a single daily document that replaces opening twelve apps.

You’ll leave with

A unified morning document combining brief, email, tasks, relationships, reading, and energy.
Fallback rules for each module in case its source is unavailable.
A consolidated approval section with all suggested actions in one place.
A voice-ready version of the informational sections.

Use the prompt in order

1
List the modules to include and their outputs
Name each and what it contributes to the morning document. The morning brief provides the executive summary and dossiers. The email provides urgent items and drafts. The task module provides today's three things. Add relationship alerts, reading highlights, and energy prompts if those modules are active.
2
Write fallback rules for each module
For each , specify what happens if its source is unavailable. The goal is a morning document that always runs, even if some sources are down. 'If email is unavailable, note it and produce the rest. If relationships are not built yet, skip that section.'
3
Collect the approval boundaries into one section
Instead of reviewing approvals per , collect them into a single list at the end. Each approval item should name the action, the module it came from, and what happens if you approve or dismiss it.
4
Add the voice layer for informational sections
Mark which sections should be narrated (summary, dossiers, tasks, flags) and which should stay text-only (approval items, detailed data). Apply your tone rule from Turn Any Into Audio.

Prompt details

Each detail will be inferred, or you will be asked to clarify in the chat.

2/2 details

2 details will be inferred, or you will be asked to clarify in the chat

Starter prompt text

Open the full text if you want to check what will be copied.

Show starter prompt

Run my morning. Combine these modules into one unified document:

Modules to include:
Infer from what you know about me or ask if it is unclear

For each module, produce its output using the sources and rules I have set up. If a module's source is unavailable, note what is missing and produce the remaining sections. Never skip the entire morning document because one module failed.

Format the document as:
1. Executive summary (5 lines: meeting count, conflicts, urgent emails, top tasks,
   one flag).
2. Per-meeting dossiers (attendees, prior context, prep suggestion). 3. Today's three tasks (matched to energy and calendar). 4. Relationship alerts (cooling contacts, birthdays, reconnection suggestions). 5. Resurfaced reading highlight (relevant to today's meetings or projects). 6. Energy check-in prompt. 7. Approval queue: all suggested actions from all modules in one list, each with
   the action, the source module, and what happens if I approve or dismiss.


Voice version: narrate sections 1 through 6 using this tone: Infer from what you know about me or ask if it is unclear. Section 7 (approvals) stays in text only.

What the answer should give you

A unified morning document with all seven sections. Executive summary shows 4 meetings, 1 conflict, 2 urgent emails, 3 top tasks. Meeting dossiers for each. Three tasks matched to energy. One relationship alert (David, 34 days silent). One resurfaced highlight about stakeholder management. Energy prompt. Approval queue with 3 items: send drafted reply, move standup, reconnect with David.