Blog

One dashboard, nine projects: 54 roles configured, workers offline

I am building a private command center that runs my nine projects from one screen, with 54 agent roles configured across the teams and a CEO bot in testing on top. Here is the honest picture of where it stands.

Actual Command Center Today dashboard
Fig. 01 · The actual Today dashboard.

The real screen

The Work page: the weekly batch is waiting for review and both workers are not registered yet

The weekly approval detail for the Project A batch: seven posts for the week of November 9, all seven passed review, zero selected so far, with the post list, the preview and the guard and reviewer checks

The search and command overlay: search jumps to a page, Tab asks Claude read only, and agent jobs still wait for approval

Friday night, November 6. Seven posts for Project A drafted and reviewed, waiting for my approval. The Work page says it plainly: approved jobs only start once a worker passes a test, and neither worker is registered yet. Nothing here posts, sends or spends on its own.

This is the actual dashboard, not a mockup. The status line says two of nine projects are active and the workers are still offline. I am sharing it because this is what the early days of a system like this really look like.

The setup

Nine project tiles reporting into one Command Center hub, each tile with its stage and six agent role dots

Six agent roles per project. Nine teams of six is 54 agent roles configured, all reporting into one screen. Two projects are active today, both run through Hermes. The other seven sit at earlier stages, from exploratory to waiting.

Each project is designed around its own team of six bots, and every team reports into the same screen. The idea is that I run nine projects at once without holding nine inboxes.

The Portfolio table: nine projects in five groups, each with its stage and next step, from active to waiting, two active

Every project with its stage, from active to waiting. Two are active this week.

The Systems page: the Main PC hosts the dashboard, the AI PC is the always on worker, and the tools and accounts table lists what is signed in, what is tested and what still needs me

Two computers run it. The main PC hosts the dashboard. The AI PC is meant to be the always on worker, and it is not registered yet, so jobs still wait for me. Below, the tools and accounts it depends on and what state each one is really in.

How a week runs

  1. This week the agents drafted seven posts for Project A.
  2. All seven passed the automatic review.
  3. The batch is now waiting for my approval.
  4. Nothing publishes on its own. Drafts only.

That is the intended loop for every project: the team prepares, review runs, and the decision comes back to me. Today one project has real output in that loop. The rest are at earlier stages.

Talk to the CEO bot

The path from my phone through the CEO bot and the project teams to a draft batch that waits for my approval

The newest piece is voice. From my phone I can talk to the CEO bot, the bot I am now testing to manage the whole dashboard as one company, and it turns the request into work for the right team.

Working today vs still in testing

Working today:

  • Seven reviewed drafts waiting for my approval
  • A posting week calendar I approve
  • System health on the board

In testing:

  • The CEO bot coordinating the teams
  • Voice control from my phone
  • Workers that run jobs on their own

One system is down this week, the Telegram alerts, and it stays on the board until it is fixed. The dashboard is built to show problems, not hide them.

For Concierge clients

For Concierge clients: an agent team set up for your business, one dashboard to see all of it, your approval at the center of every action, with the status checked November 6, 2026

This is what I can configure for Concierge clients: an agent team set up for your business, one dashboard to see all of it, and your approval at the center of every action. You stay the owner of every decision. The agents do the preparation.

Status checked November 6, 2026. Two of nine projects active, 54 agent roles configured, workers offline while the CEO bot is in testing.