One organized home for your projects, your AI tools, and what needs to happen next. Hand work between AI tools, keep project context in one place, and see your next useful action without digging through scattered chats and task lists.
Get the Guided Setup See it in action
You've got Claude in one tab.
ChatGPT in another.
Maybe Codex building something. Maybe Grok open on X.
Each one is brilliant. None of them knows what the others are doing. None of them remembers yesterday.
So your projects live in ten chats, your tasks live in your head, and you spend half your day re-explaining your own business to software you already pay for.
The tools aren't the problem. The missing piece is a shared source of truth and a line that feeds them work.
You sit down, open one dashboard, and press one button: DISPATCH NEXT TASK. It looks across every project and hands you your next useful action, weighing priorities, deadlines, dependencies, and the capacity you actually have today. Your work and agent work stay separated, so an AI's to-do list never crowds out your own next step.
Your context lives in files, not in chat history. The dashboard and file-capable agents work from the same task file: when an agent saves a result and moves its card to Review, you see it appear without exporting or importing anything. Browser AI joins in through a handoff brief you paste in.
I spent over 100 days building an AI headquarters for my own life and businesses. Most of what I built turned out to be bloat. This is the simple assembly line that survived, cleaned up so you can run it yourself.
Your context, projects, decisions, and inbox live as plain files. Local, or synced through OneDrive, Dropbox, or Google Drive. You own everything.
A local dashboard connected to your shared board file. File-capable agents update cards; you review their outputs. Your work moves through Intake, Queued, Ready, In Production, Review, Complete.
One button picks the next task using urgency, importance, due dates, dependencies, revenue impact, and your capacity today. It shows you why.
Claude and other file-capable tools read the folder directly. ChatGPT and Grok get an auto-generated handoff brief. Nothing gets re-explained twice.
Watch a task travel. Human review is built in: work returns to you for review; publishing is a separate action. You stay the editor-in-chief; the machine does the hauling.
The full walkthrough: my desk, the real system, one honest take.
Real screenshots from the system you download. The example data ships with it so you learn by clicking.
Stop deciding what to work on. One press looks across ALL your projects and hands you your next human action, with the reasoning printed right on the card. Viewing one project never hides more important work sitting in another.

Heavy, medium, or light: you tell the system your honest capacity and it plans around the day you actually have. I built this living with chronic pain. A system that only works on your best day is not a system.

"Your Next Action" shows what needs you. "Agent Work" shows the separate queue an AI can handle. Assigning a task to an agent parks it in their lane; it never crowds out your own next step, and it doesn't secretly launch anything.

Each project gets its own tab and the same clear line: Intake, Queued, Ready, In Production, Review, Complete. Tabs filter your view; Dispatch still watches everything, so focusing on one project never blinds you to the rest.

ChatGPT and Grok can't read your files. Nobody's tools magically share memory, whatever the gurus say. So the dashboard generates a complete context brief you paste into any browser AI: who you are, the project, the task, what done looks like.

You don't type your projects in by hand. The kit includes interview prompts: paste one into your AI, answer its questions honestly, and it builds your entire board: projects, tasks, owners, priorities, due dates.
No subscription. No account. Your board and files stay in your own folder. Full shared mode needs free Python 3.10+, a browser, and a file-capable AI tool you already use; AI subscriptions and usage are separate.
The DIY kit works. But the fastest version of this is the two of us on a screen share wiring it into your actual life.
Before the call you fill out a short questionnaire. I build your starting configuration before we ever meet: your storage choice, your terminology, your AI tools mapped to roles.
Then 90 minutes together on Windows or Mac: get the dashboard running, connect your AI, and learn the workflow using your first real project. We allow about 30 minutes for setup, 45 for hands-on use, and 15 for questions and a final check. We adjust that split to your computer and needs; setup time varies.
You get a short installation-fix window after the call in case anything in the setup misbehaves on your machine. It's a setup service, not unlimited coaching; deeper work and future calls are separate and I'll always be straight with you about which is which.
Complex or fully done-for-you builds get a custom quote after the application. I'd rather tell you the real scope up front than surprise you later.
No, and nobody's system does that. An AI app with verified access can work from the shared folder. A browser chat does not automatically see your local files. Paste the dashboard's handoff brief and supply the relevant project text or attachments. The brief contains a board summary and shared context, not every project file.
The kit includes step-by-step Windows and Mac guides plus an AI-assisted setup interview. Full shared mode needs Python 3.10+ and a local launcher; Mac setup may use Terminal. A browser-only preview is also included, but it does not share live cards with agents. Guided Setup is for people who want help getting the real workflow running. Your AI apps and usage are separate.
Both. Windows with Chrome is what I use daily. I also completed the core workflow on an M2 Mac running Sequoia 15.6.1 in Chrome and Safari: saving, backups, AI updates, review and approval. On that Mac, Apple blocked the shortcut, so I used the included Terminal instructions after installing Python. The clear quick-start PDF covers both systems. Other computer configurations may differ. OneDrive, Dropbox, and Google Drive can sync the folder; use one active computer at a time.
A new AI model does not automatically mean rebuilding your Assembly Line. Changes to an app's file access or workflow may need an adjustment. Fixes and updates I release for the version you bought are included. That is not a promise of perpetual compatibility or a fixed update schedule. If a separate major upgrade is offered, its price will be clear and buying it will be optional. Later hands-on reconfiguration is a paid working session, agreed before we start.
Yes. We can agree on a later setup date before you purchase, and an early-bird purchase locks that price for the agreed date. The offer is for the first five purchases, subject to actual availability. Contact me first if you need to plan further ahead. It is not an indefinite reservation. The seven-day installation-fix window starts after our call.
No, and be suspicious of anyone who claims theirs does. Assigning a task to an agent puts it in that agent's queue; you start the agent in its own app with the included worker prompt, or set up a bounded recurring job there using the included guidance. The dashboard is not a hosted automation service, doesn't call AI services itself, and nothing runs with your computer off. What IS automatic: when a file-capable agent finishes and moves its card, your open dashboard shows it without any exporting or importing.
Any of them. It ships configured for Claude, Codex, ChatGPT, and Grok, and the config file lets you add or rename tools in one place. The system doesn't care which AI wins next year; it's files and a browser page.
No. One-time purchase. The system is a folder you own. There's nothing for me to shut off.
The dashboard server listens only on your computer and stores cards in your own board.json file. Project files stay in your chosen local or synced folder. Your AI tools and sync provider have their own data policies, so grant access deliberately. No product analytics are included, and setup does not require sharing passwords or API keys with me.
Every label, stage, capacity word, and color is customizable from one config file. If you buy it and it truly isn't for you, reply to your receipt within 30 days and I'll refund the DIY kit, and if you'd rather fix it than refund it, I'll hop on and help. Guided Setup refunds any time before the call happens; after the call, the work is done and it's yours.
It's my brand. Bedridden in 2014, won a 200-mile ultra later, chronic pain every day since. I build systems that work on bad days, because I don't get to have only good ones. That philosophy is baked into this: capacity is a first-class input, not an afterthought.
If you use more than one AI and you're tired of being their only shared memory, this is the fix. Pick your lane above, or start with the one most people should: the Guided Personal Setup.
Get the Guided Setup