before we build together
Your AI COO, on the machine you own.
A short setup list that makes your install smooth and your system truly yours. Most items take five minutes. Your setup session is about an hour once these are in place.
ownership, clearly stated
You own it.
No lock-in.
No lock-in.
The system and all your files live on your Mac. You are not renting a platform. You are not locked into any vendor, including any single AI provider. You bring your own AI account and your own code repository, and the system runs on your hardware, on your terms.
Your Mac, your files
Everything runs on your hardware. Your data, your notes, your AI's memory: all of it lives on your machine and stays there. We never hold a copy.
Your choice of AI, not ours
Claude is the recommended default and the simplest path. But you are not locked into any single AI provider. You can run a local model on your own machine instead. The AI account is yours, the usage is yours, and you are never locked into one vendor.
Your own private GitHub repository
Your private GitHub repo is simply how your install lives and how you receive improvements. The repo is in your GitHub account. You own the code and the history, forever.
Optional cloud connectors you control
Optional integrations (Gmail, Calendar, Zapier) let the system reach your other tools. You authorize them with your own accounts. You can revoke any connection at any time. The core system runs without them.
The plain truth: you own your files, your system, and your Mac. Revoke our access at any moment and your system keeps running, independently, with no interruption.
before we build your AI COO
What you need
in place, before we start.
Three groups. Most people have half of these already. Group A is required; Groups B and C are optional.
Group A
Required
Please complete these before your build session.
Your Mac
Apple Silicon (M1 or newer) with at least 16 GB of RAM. This is the machine your AI COO will live on.
AI Account (Claude recommended, or local model)
Recommended: an active Claude subscription (Max plan is the simplest path). Claude is the default AI the system runs on, and your account means your usage, your history, and your data belong entirely to you. Prefer to keep everything local? You can also run a supported local model on your own machine. You are not locked into any single AI provider.
Obsidian
Download and open the free app. This is where you will read, review, and edit your AI COO's work in plain text you always own.
obsidian.md
GitHub Account
Create a free account, enable two-factor authentication, keep your repository private, and send us your username. This is how you own your install outright and receive every future update.
github.com
Your Skill Library
Download your full capability library to your Mac. This gives your AI COO its complete operating toolkit from day one.
Connected Accounts
Connect Gmail, Google Calendar, and Zapier through Claude's cloud connectors. Zapier links your other tools. This lets your AI COO work across everything you already use.
Your Wish List
Write down what you wish your operating system could handle for you. We design your build with you, and this shapes everything. Nothing is too small or too ambitious to include.
Group B
Recommended
Optional but worthwhile. Each is free and takes minutes to set up on your own.
Privacy and Security Layer (Objective-See)
Three free tools that put you in control of your Mac: LuLu shows you outbound connections, KnockKnock scans for anything that persists, and BlockBlock stops new persistent installs. All free.
objective-see.org
Amphetamine
A free Mac App Store utility that keeps your machine awake so your AI COO can run around the clock without interruption.
apps.apple.com → Amphetamine
VoiceInk
Lets you speak directly to your AI COO by voice, anywhere on your Mac. Optional, but a favorite for fast morning check-ins.
tryvoiceink.com
Group C
If We Manage Your Mac
Only needed if you choose remote-assisted setup or ongoing managed support.
Tailscale
A private, encrypted connection that lets us set up and maintain your install remotely. You own the network and grant us scoped, revocable access on your terms. We walk you through the setup, and you can revoke access anytime.
tailscale.com
obsidian setup
Set up Obsidian
like ours.
A clean Obsidian install takes about ten minutes. This is the exact configuration we use: the theme, the plugins, and the core features that power your CRM, task system, and daily notes. We dial in anything you miss together during your setup session.
step one
Install the
Minimal theme.
Minimal theme.
Open Settings, go to Appearance, and search for "Minimal" in the community themes. Install it, apply it, and switch to light mode. Set your base font size to 16. The result is a clean, distraction-free workspace that reads like a well-designed document.
Theme: Minimal
Color mode: Light
Base font size: 16
We help you dial this in. You do not need to get every setting perfect on your own. During your setup session we configure Obsidian together and make sure everything is working exactly as it should before we finish.
step two
community plugins
Four plugins to install
Settings → Community plugins → Browse
Minimal Theme Settings
Fine-grain controls for the Minimal theme: card view, image grids, table styles, and more.
Folder Notes
Attaches a landing note to each folder so your top-level areas open to a summary, not an empty list.
Recent Files
A sidebar panel showing your most recently opened notes. Lets you jump back to active work in one click.
Better Word Count
Accurate writing stats in the status bar, including selection counts and session word totals.
core plugins
Nine features to turn on
Settings → Core plugins
Daily Notes
Templates
Properties
Bases
Graph View
Backlinks
Command Palette
Bookmarks
Outline
Bases powers the database views your CRM and task system run on. Properties and Templates keep your notes consistent. The others are standard navigation you will use every day.
how it works
Installed on your machine,
owned by you.
A white-glove setup on your own Mac, on your own Claude login. You own everything. We build it, wire it, and maintain it with you.
your hardware
your Mac, you own it
The AI COO runs locally on your Mac. You log in with your own AI account (Claude by default, or a local model if you prefer). Your data and your keys never leave your machine. No cloud middleman, no platform to depend on.
your codebase
your private GitHub repo
Your install is a private GitHub repo, custom-built for your business. Your AI COO keeps getting sharper: we ship refinements, your install pulls them in, and nothing you customized is disturbed. The repo is yours, in your GitHub account, forever.
our side
our remote care
We set up and maintain your install over a private, encrypted connection. We use it to configure, test, update, and fix. You can revoke that access at any time and everything keeps running.
revocable by you anytime
the installation process
Eight steps, fully documented.
step 01
kickoff and intake
we learn your business
We capture your business, your clients, your tools, and what your system should do. This intake becomes your custom config: the specialists we tune, the data sources we connect, and the outputs we build for your operation.
Happens before we write a single line of code.
step 02
we build your install
your private repo, custom built
We engineer your private GitHub repo from the ground up: your agents, your skills, your connector configs, all built for your operation. The custom work happens here, on our side, before we ever touch your Mac.
You can inspect every file in the repo at any time.
step 03
one setup session
about an hour, screen-share
You grant us access once via a screen-share. We install the runtime, pull in your private repo, and configure the schedule directly on your Mac. This is the one hands-on session on your machine. Total time: about an hour.
After this session, remote access is how we reach you going forward.
step 04
you connect your accounts
your logins, your keys
You log into Claude Code with your own AI account (Claude by default, or a local model if you prefer). You authorize your tools with your own credentials. Your keys go into your Mac's encrypted Keychain. We never use our accounts on your machine.
We guide the steps. You enter the credentials. Always.
step 05
your connectors get wired
data sources, secured
We connect your data sources through secure per-client API keys stored in your Keychain. Each key is scoped to the tool that needs it: read only, never shared, never visible to the AI model, never stored in your GitHub repo.
One key per connection. Revoking one never affects the others.
step 06
scheduled work runs on cadence
on your schedule, on your Mac
A scheduled job on your Mac fires your configured tasks on the clock: daily briefings, weekly content, pipeline reviews, and more. Outputs land where you already work. You approve what goes out. Nothing publishes without you.
Powered by macOS launchd. No third-party scheduler, no cloud cron.
step 07
we maintain it remotely
improvements via GitHub
We push improvements through GitHub: your repo pulls the update and it runs on the next schedule. For fixes or extensions you request, we connect over the private remote session, work, and disconnect.
All code changes go through the repo. Nothing happens on your machine without a commit trail.
step 08
you own it, forever
the end state
Your Mac, your repo, your AI account, your Keychain, your data. Revoke our remote access at any time: the system keeps running on schedule without us. It is yours to run, extend, or hand to your team. No vendor lock-in.
This is what "you own it" means in practice.
your own AI account runs it
The AI COO uses your own AI account (Claude by default, or a local model if you prefer), not ours. We never put our credentials on your machine. You control the account, the usage, and the data that goes through the model. You are not locked into any single AI provider.
your Mac, on your terms
For on-demand use, your Mac only needs to be on when you are using it. If you want continuous, always-on automation, including scheduled work and background systems that run without you, keep it on. A dedicated Mac Mini is the clean always-on host. Amphetamine keeps a laptop awake. Your choice.
your data stays on your machine
It lives on your Mac. Only the slice a task needs goes to the AI model, which never trains on it and does not store it. We have no copy, no database of your business data, no ongoing access beyond what an active remote session lets us see.
what you are getting
The engine
working for you.
A full team of specialists, directed by one AI COO, reading from one shared brain that knows your business and gets smarter every single day. You never explain yourself twice.
your specialist team
Content
Drafts your LinkedIn, Substack, and long-form content in your voice, consistently and on schedule.
Sales
Manages your pipeline, preps your calls, and keeps your CRM current so no opportunity slips.
Operations
Handles your inbox triage, task review, calendar prep, and daily briefing so every morning starts clear.
Design
Creates branded visuals, carousels, and graphics from your content so you always look the part.
Research
Deep-dives on prospects, competitors, and industry moves, surfacing what you need before the call.
Growth
Monitors what is working, finds the levers, and keeps your pipeline engine running between sessions.
Product
Helps you package, price, and position your offers so your expertise scales beyond your hours.
Client Value
Keeps a pulse on every client relationship and surfaces proactive ways to deepen the value you deliver.
the shared brain
One brain.
Every specialist.
Every specialist.
Every specialist reads from and writes to a single shared brain that holds your business context, your voice, your decisions, your client history, and your goals. It compounds over time. The longer you use it, the more it knows, and the less you have to manage.
Knows your clients, your voice, and your current priorities
Remembers every decision so you never re-explain context
Gets smarter every session, automatically
Your data lives on your machine, owned entirely by you
the AI COO
One point of
command.
command.
Your AI COO sits above the team: directing, delegating, and synthesizing. You give it one instruction and it coordinates the rest. No juggling specialists. No managing tools. Just outcomes.
1
You direct. Your AI COO proposes. You decide. The human stays in control of every consequential call.
2
It runs. Between sessions, your AI COO keeps priorities moving, surfaces what changed, and flags what needs you.
3
It compounds. Every session adds context. Every decision is remembered. The system gets better without extra effort from you.
how it evolves
From session one
to run and grow at the speed of your thoughts.
Your AI COO does not need months of configuration. It starts useful on day one and deepens fast.
Day 1
Learns your voice
Your AI COO reads your existing content, your past decisions, and your goals. From the first session, it writes in your voice and works within your context.
Week 1
Knows your business
It has seen your pipeline, your clients, and your priorities. It stops asking basic questions and starts making meaningful connections between what you are working on.
Month 1 and beyond
Run and grow at the speed of your thoughts
It surfaces the right things before you ask. It flags risks you have not thought about yet. It runs the operational layer so you can stay focused on the work only you can do.
your role
What you do.
Your role in this system is intentionally light. The work that used to fill your administrative hours now happens in the background. Your attention goes to decisions, relationships, and the work only you can do.
during onboarding
Log in, run one command, authorize your tools
During the setup session, you log into Claude Code with your own AI account (Claude by default, or a local model if you prefer), run a single terminal command to set up scheduled authentication, and authorize your integrations through your normal browser login flow. You see exactly what access is requested before you approve. That is the full technical lift on your side.
day to day
Review what arrives. Give direction when you want.
Reports, summaries, and drafts land where you already work: your inbox, your notes, your project tool. You review them. If something is ready to send, you send it. If something needs a change, you note it.
- Nothing goes out without your review unless you pre-approve a category
- Every output shows what data it drew from
- Open Claude Code any time for a direct, ad-hoc request
- Your context is already loaded: no re-explaining required
what you do not do
You do not manage the system
You do not monitor logs, restart jobs, or update configurations yourself unless you want to. That is our ongoing responsibility. Your Mac should feel like it has a quiet operator in the background, not like a piece of software that needs tending.
how to talk to it
Give it direction. It drafts. You approve.
Your system understands plain language. You do not need to learn a command syntax or navigate a UI. Describe what you want, and it uses your context to do it well.
client context
Using what the system already knows about your clients
"Pull together everything from the last two months with this client and give me a summary of where we are."
The system has your client configuration from intake. It knows who they are, what they engaged you for, and what outputs you have produced. Client-specific requests use this context without you having to re-explain it.
report requests
Pulling a report outside the normal schedule
"Give me the weekly pipeline summary early, I have a call this afternoon."
Scheduled reports run on their calendar. But you can request any configured report ad hoc at any time. It pulls the same data sources and formats it the same way. You do not have to wait for the schedule.
drafting
Drafting communications in your voice
"Draft a follow-up to this client after today's call. We agreed to move forward and they want a proposal by Friday."
Because your system is configured with your voice, your tone, and your communication patterns, drafts feel like drafts you would write. You review and send, or adjust and send. You always have final say.
business intelligence
Making sense of what is in front of you
"What does my pipeline look like right now, and what should I be paying attention to this week?"
Your system reads your calendar, your CRM data, and your recent outputs to give you a current-state view and a prioritized focus. This is the kind of question a chief of staff would answer without you having to pull the data yourself.
how your AI COO works for you
A full team running
your operations.
Reports are one thing your AI COO produces. It also drafts your content, manages your pipeline, preps your calls, handles research, runs your task system, and more. Everything lands where you already work. You approve. Nothing goes out without you.
content, sales, and operations
A full operations layer, not just reports
Your AI COO ships 10 specialists and 31 skills covering content creation, sales and CRM, operations, marketing, SEO, design, meeting prep, calendar and task management, research, drafting, and weekly planning. Reports are one output among many. The system runs the whole operational layer so you stay focused on the work only you can do.
- Content: drafts, schedules, and refines in your voice
- Sales: pipeline management, call prep, CRM, follow-ups
- Operations: inbox triage, task review, daily briefing, calendar
- Research, design, SEO, and more, on demand or on schedule
client-facing drafts
Client updates arrive as drafts, never sent automatically
If your configuration includes client-facing outputs (status updates, weekly recaps, follow-up notes), these arrive as drafts in your inbox or project tool. You review and send. The system does not send client-facing communications on your behalf without your explicit review, because your relationships are yours.
- Always arrive as drafts, never sent automatically
- Written in your voice, based on your approved formats
- Reference the specific client context from your onboarding configuration
scheduled outputs
Scheduled work runs automatically on your Mac
Scheduled jobs run at the frequency you specify during setup. They pull from your connected data sources, compile into your approved format, and deliver to your configured destination: your inbox, your project folder, your notes. There is nothing to trigger and nothing to check.
ad-hoc requests
On-demand outputs from direct requests
When you give your system a direct request, it uses your full context: your client history, your past outputs, your preferred formats. The output reflects your actual business. Request a report, a draft, a research brief, a pipeline summary, or any other task at any time, on demand.
always true
Four things that never change
You approve what goes out
No communication goes to clients or external parties without your review. Drafts arrive for your judgment. Nothing sends automatically unless you have explicitly pre-approved a destination.
You can always reach us
If something does not look right, something is missing, or you want to change how the system behaves, you reach out and we respond. This is a managed engagement, not a product you operate alone.
Your accounts stay yours
We never hold your login credentials, your API keys, or your account passwords. All access is through OAuth flows you authorized, stored on your hardware, revocable by you at any time.
Your Mac, on your schedule
The system runs on your Mac. For on-demand use, it only needs to be on when you use it. For continuous background automation, keep it on. You decide the mode. You have complete control over when it operates.
ownership and control
Everything you own.
Revoke anytime.
This is not a hosted service where your data lives on someone else's server and your access depends on a subscription to a platform you do not control. Every component of your AI COO belongs to you.
What is yours, completely
Your Mac
The system runs on hardware you own. We access it only for maintenance, through a secure connection you control.
Your AI account (your choice)
Claude is the recommended default. You log in with your own subscription. Or run a local model if you prefer. Either way, the AI account is yours, usage is yours, and we never hold your credentials.
Your GitHub repository
Your configuration lives in a private repository in your own GitHub account. You own the code and the history.
Your API authorizations
Every tool integration is authorized through your own OAuth flows. Tokens live in your Mac's Keychain. You can revoke any authorization at any time without our involvement.
Your data
Client data, report outputs, and conversation history stay on your machine. Nothing is sent to an external server we operate.
Your network access
You control the private network that allows us to access your machine for maintenance. You can revoke our access in seconds, any time, for any reason.
Your configuration
The settings, report formats, and details that define how your system behaves are documented in your repository. They are readable and editable by you.
Your outputs
Every report, draft, and summary produced by your system belongs to you. There is no platform license on your own AI's outputs.
if you want to leave
What "revoke anytime" actually means
You can end this engagement at any point. To remove our access to your machine, you revoke the network key you issued us. That is it. We are out immediately. Your system continues to run on your hardware, with your configuration, with your accounts, indefinitely. You take everything with you and can operate it independently.
if you want to update something
How changes and improvements work
Configuration changes (new clients, new report formats, new integrations) come through us: you describe what you want, we update the configuration and deploy it. Engine improvements ship to your system via your repository and we apply them during a maintenance window. You are notified before anything changes.
questions answered
Frequently asked questions.
The questions we hear most, answered plainly.
Yes. It runs on your Mac, from your private GitHub repository, on your own AI account. You own the machine, the repo, your data, and every credential. You can revoke our access or shut it off anytime.
Your data stays on your Mac. It is never stored in your GitHub repository, which holds code only. The only thing that leaves is the slice a report needs to send to the AI model, which never trains on it or keeps it.
Claude is the recommended default and the simplest path to get started. You log in with your own subscription and the account is entirely yours. But you are not locked in: you can also run a supported local model on your own machine if you prefer to keep everything fully local. We never put our account on your machine.
A Mac with Apple Silicon and at least 16 GB of RAM, an AI account (Claude recommended, or a local model if you prefer), Obsidian, a private GitHub account with two-factor authentication, and your wish list. The full checklist is above.
About an hour together on a screen-share, once your prerequisites are in place. We build it with you.
Not always. For on-demand use, it only needs to be on when you are using it. If you want continuous, always-on automation (scheduled reports, background systems, ongoing tasks running without you), then keep it on. A dedicated Mac Mini is the clean host for that mode, and Amphetamine keeps a laptop awake if that is what you are using. You choose the setup that fits how you want to work.
We ship refinements through GitHub and your install pulls them in. Your custom work is never disturbed, and we never touch your machine to update it.
Only if you choose the managed option, and only over a private connection that you own and control, scoped to what we need. You can revoke our access anytime from your side, and it is final.
Your code is safe in GitHub. We keep a documented recovery process to rebuild your install and re-enter your keys.
It runs your operations. Reports, drafts, research, and tasks across the tools you connect, directed by your AI COO and a shared brain that learns your business. You approve anything that goes out.
what happens next
Once your prerequisites
are in place, we build together.
Once you have Group A checked off, we book your setup session. About an hour together, and we build your AI COO from the ground up. Your data stays on your machine. Your credentials stay in your Keychain. You own it the moment we finish.
Book your setup session →1
Complete your prerequisites
Group A is required. Group B and C are optional and can follow after your initial build.
2
Share your GitHub username
We use it to provision your private repository before the setup session begins.
3
Book your setup session
About an hour on a screen-share. By the end, your AI COO is live and running on your machine.
4
Run and grow
From day one it learns your voice and context. Week by week it deepens. You own the system, forever.