Build a Personal AI Command Center in 30 Minutes

A free step-by-step guide for ambitious leaders. Stand up a personal AI Chief of Staff that knows your goals, calendar, and relationships. Four steps. No engineering background needed.

Why this matters

Most people use AI like a search box. Type, answer, close tab. A personal Command Center is different. Your AI knows your goals, your calendar, your relationships, your week. It writes you a morning brief that says what actually matters today.

Models are getting smarter every quarter. The people getting outsized returns aren't the ones prompting better. They're the ones who built a system that lets more AI capability land on the things they care about. You're shipping an MVP, not a perfect product. 30 minutes from now, something is running. You iterate from there.

The 30-Minute MVP

Four steps to a working Command Center

Steps 1 through 3 are your MVP. 30 minutes, end to end. Step 4 is the rest of your life with the system, getting sharper every week. Each step gives you a prompt to paste into Claude Code. Claude Code handles the setup. You bring the context.

Personal first. This guide is for your personal Command Center, the one that knows your goals, relationships, and life. Many leaders eventually build a separate work Command Center on a different machine, with its own rules and a hard firewall. Start with personal. The pattern transfers.
1

Tell it who you are

A Command Center is only as good as what it knows about you. Start here.

The starter set

Gather these into a folder. Don't worry about formatting. Claude Code sorts it.

That's enough. This alone changes every conversation you have with your AI from here on.

Paste into Claude Code
I want to start building my personal Command Center. I'm going to give you the basics about me, and I want you to organize them into a context folder you can read at the start of every session.

Create a folder called ~/CommandCenter/ and inside it create a MEMORY.md file. Then walk me through what to drop in. I'll have my resume, my LinkedIn, a one-pager about me, a few writing samples, and my professional bio.

Once we have the basics, set up a session-start hook so this context loads automatically every time I open a Claude Code session. I shouldn't have to re-explain who I am.
Want more signal?

Optional sources for a sharper system

  • Your calendar. Today plus the next 30 days.
  • Meeting transcripts. Granola, Otter, or Fireflies.
  • Health data. Oura, Whoop, Garmin.
  • A simple contact list. Even a CSV of who matters and how you know them.
  • Private writing. Journal entries, draft posts you never shipped, voice memos.
Every source is a choice, not a requirement. The system runs on your machine. Nothing leaves it unless you wire it to. The tradeoff between signal and what you share is yours to make.
2

Tell it how to talk to you

A great brief in a place you never read is worth nothing. Tell your AI how you work before you build anything.

Pro tip

Turn on Claude Code voice mode

Before you start, enable voice mode in Claude Code and answer these out loud. You'll move twice as fast, and the answers will be more honest. Less editing, more truth.

When should it reach you?

How should it sound?

What do you actually want from it?

Paste into Claude Code
I want to define how my Command Center should communicate with me. Walk me through the questions in Step 2 of the guide, one section at a time. As I answer (I'll probably use voice mode), take notes in ~/CommandCenter/MEMORY.md under a section called "How to work with me."

Then write a one-paragraph voice and cadence summary at the top of that section. This is what every future brief should follow.
A worth-it detour

Write a one-page vision

Write a one-page picture of your life one year from today, in present tense, as if it's already that day. Where you live. What you're working on. What a Tuesday looks like. What you stopped doing.

Mine sits at the top of my Command Center. Every real decision maps back to whether it gets me closer to that document. Talk it into a voice recorder. Describe the day.

3

Have it write your daily brief

One prompt. Claude Code sets up the rest.

Every morning at the time you chose, an email lands in your inbox. It reads your context and your calendar, and tells you what matters today, what to drop, and what to watch out for. Paste the prompt below and answer the follow-ups.

Paste into Claude Code
Set up my daily morning brief.

The brief should fire at the time I prefer (ask me) and land in my inbox as an email. It should read my Command Center context (the MEMORY.md and the one-page vision I wrote), my calendar for today and the week ahead, and any recent activity logs. It should then synthesize a 200 to 400 word brief that names:

- What today actually is (calendar shape, energy state, what's load-bearing)
- One or two things that matter most
- What to drop
- What pattern to watch out for in myself

Use my voice preferences from MEMORY.md. No corporate hedging. No prescribed counts.

To make this work, you'll need to:
1. Set up a daily scheduled job on my Mac
2. Stand up a small local server that reads my context
3. Wire the brief into my email

Walk me through each step. Pause for my input when you need information from me (calendar access, email setup). Make it run by tomorrow morning.

Claude Code will ask a few things along the way: calendar access, where to send the email, your preferred time. Answer each one.

What you'll have when this finishes
  • A scheduled job that fires every morning at your chosen time
  • A local context engine that reads your goals, calendar, and activity
  • An email that lands in your inbox before you've had coffee
  • Pennies a day to run, dropping
The first week the brief will be slightly wrong. Tell it. The system learns.
4

Iterate from there

Your MVP is running. Now you make it sharper. A Command Center compounds because you keep teaching it.

Three practices to build in

Paste into Claude Code
Audit my Command Center. Tell me:

- Which scheduled jobs are firing, and which are stale or silent
- Which briefings are actually reaching me and which I'm ignoring
- What context in MEMORY.md is out of date
- Three small improvements I could make this week

Be direct. Don't pad. I'd rather hear what's broken than what's working.

Add new surfaces when they earn the right

Once your daily brief is solid, you can add a weekly relationship digest, a pre-meeting briefing, a content-prompt that reads your last month of activity. None of these need to exist on day one. They earn their place when the daily brief is working.

When you're ready

Adding a work Command Center

Most leaders eventually build a second Command Center for work. Different machine, different rules, hard firewall. It holds client templates, brand styling, organizational context, the language of your team. Same pattern as this guide, different context. Build the personal one first.

What comes next

Your MVP works. Now the real leverage. Three directions for your next stage, in roughly the order most people take them.

Stage 2

Connect your real data

A Command Center gets exponentially more useful when it can read from the places your life already lives. Most of these connect through the command line or a small local server that runs on your machine.

  • Google Workspace. Calendar, Gmail, Drive, Docs. The single biggest unlock for most people.
  • Meeting transcripts. Granola, Otter, Fireflies. Past conversations become searchable.
  • Your CRM or contact list. Relationship intelligence on top of warmth signals.
  • Health data. Oura, Whoop, Garmin. The brief factors in how you're recovering.
Stage 3

Lock down the basics

Once real data flows, get the security right. Five rules cover most of what matters.

  • One credentials file, never in git. Keep API keys in a single locked-down file on your machine. Add it to your gitignore on day one.
  • Local server, loopback only. Your Command Center's brain should never be reachable from the public internet.
  • Rotate when in doubt. If a key was ever in a shared place, rotate it. Cheap insurance.
  • Separate personal and work keys. Never reuse a credential across boundaries.
  • Read before you connect. Whenever you add a new integration, read what permissions you're granting. Skim it. Ten seconds. Saves real pain.
Stage 4

Add more surfaces

The same brain can power more than the morning email. The pattern is identical: a scheduled trigger, a synthesis call, a place the answer lands.

  • A weekly synthesis. What changed this week across your goals, network, body, pipeline.
  • Pre-meeting briefings. Context on whoever you're seeing, ten minutes before you walk in.
  • A content prompt. Reads your last month of activity, suggests what to write about.
  • Voice mode, always. Conversational access to the same brain from anywhere on your machine.
A deeper course is in the works

Building Your Command Center, the full course

Each of the stages above is becoming its own module: connections, the brain, security, custom surfaces. If you want the full course when it ships, drop me a line.

Want help building yours?

If you want a partner to ship your MVP in one working session, I do this with leaders one-on-one. You leave with something running.

See how it works