BlogRocket Routine OS: The Whole Picture
Rocket Routine OS: The Whole Picture
Talk to your company. It answers, it gets the work moving, and you decide. Rocket Routine OS at a glance: the agentic operating system with AI agents.
In short: Rocket Routine OS is an agentic operating system for your company, with an AI leadership team of AI agents built in. Yearly plan, goals, numbers, meetings and work come together in one place, with an AI head for each of ten areas. AI proposes, you decide.
A company has a plan, and it gets lost between strategy, goals and work. Often there is no real strategy. Sometimes there is one, but hardly anyone knows it. Anyone who talks to their AI alone today also ends up with many separate chats, agents and results that are hard to bring back together. Rocket Routine OS keeps plan, knowledge and work in one place. This article walks through the whole product once, from the first chat to year-end. Each building block gets its own article after this one.
Talk to your company
Talk to your company. It answers, it gets the work moving, and you decide. The chat is the centre of the product: you talk to the Navigator, your everyday partner for the whole company, and to the AI heads of your areas. An AI head is the AI agent of an area and its VP (Vice President), meaning its lead. It is named VP plus area, for example VP Marketing. Whatever the menus can do, you can also do in the chat. The menus stay for the overview and for hands-on work.
An example: you write "Track that I call Isabel on 15 November." The Navigator enters you as the person responsible and 15 November as the due date, because you said both. It asks only for what it cannot work out, for example which area the call belongs to. It never sets a date by itself. Then it opens the form already filled in, with the missing field marked. You complete it in the chat ("That belongs to Sales") and save. The result is a work item, a single package of work with an area and an owner. When an agent needs several details at once, it shows a small question card. If the timing is wrong, you push it to your Inbox with "Later", the place where you approve things and receive reports.
If you do not know how to continue, or cannot find where to do something, just ask your Navigator. If you need a new specialist, tell the Navigator and describe it. In a short interview, the Navigator asks what the agent has to be able to do, creates it and adds it to your fleet of agents.
Not a solo system: you chat together
Personal assistants belong to one person and work with local knowledge. A company needs a shared memory, so Rocket Routine OS shares one knowledge base and one plan across the whole company. Any number of colleagues work in one chat together with several AI heads, specialists and advisers. A Team is a saved group of specialists and advisers with which you start a chat in one click.
Everyone sees the same plan and numbers. Each person reads in their own language, English or German, and their own date and number format. If you write "84,000", the Navigator shows how it read the number before saving. You can use the system on your own, and it works as a team system: your team and your AI team sit in the same chat.
Real work: from the price list to the large system
Beyond the small actions of a day, you do a company's real work with your AI team. You ask, for example, for a price list, for a complete landing page with text, images and layout, for a social-media automation in which a routine prepares posts for LinkedIn, Instagram, YouTube and X as drafts and every post waits for your approval, or for a project run with outcomes, work items and an ongoing report. Software is built this way too, not only a small tool but large, powerful systems: the VP Engineering leads the development, developer and tester agents write and test the code, end to end, with you in charge.
The agents produce artifacts, results you can open and share: reports, charts, presentations, web pages, documents and code. They use the browser, research the web and operate on-screen interfaces. The area's VP brings in the specialists a delivery needs and asks for your decision at the points that matter. On top of that comes almost everything else that happens at a computer in a company.
Any tool can be connected, and today there are about two dozen connections, for example to Google Workspace, Microsoft 365, Slack, Notion, Atlassian, Trello, Linear, ClickUp and GitHub.
The AI team and its knowledge
There are ten areas: Strategy, Marketing, Sales, Product, Engineering, Customer, Finance & Legal, People, Administration and Quality & Excellence. All ten are there from day one. Every area has a VP and a responsible person. By default the VP is the AI head and the responsible person is a human. A company can also make a person the VP of an area, for example its former head of sales. The AI head then stays as her assistant in the area, keeps its skills and specialists, and passes anything that needs a decision to the person.
The Virtual CEO keeps priorities across all areas in view and supports you. Below the VPs work specialists with clear roles, such as developer, tester or content marketer, and you can talk to them directly.
Behind the chats sits a knowledge system. Your documents are indexed and searchable, and every agent names its source. You mark important documents as a source of truth, and all AI heads then treat them as binding. About 30 knowledge personas (advisers) are there from the start, for example modelled on Peter Drucker, Warren Buffett or Daniel Kahneman. They know your whole company, advise in meetings and in the chat, and make no decisions. You add your own advisers by describing them or uploading texts. If you upload the transcript of a conversation, an agent pulls out decisions, work items, facts and open questions.
Meetings from the Daily to the yearly meeting
The Navigator opens every meeting by itself, greets everyone and asks whether to add or remove topics and invite more people. Then it leads through the agenda with guiding questions. All meetings are switched on from the start, and you switch off what you do not need. At the end of each topic it offers a work item if someone agreed to do something, and asks for the owner and the due date.
The Daily is the short check during the day: what changed, what is blocked, what is waiting for a person? In the Weekly the Navigator shows which metrics crossed a warning line, with latest value, trend and traffic light, and names missing values. It charts whether the weekly figures point to a problem and asks whether the quality matches what was agreed. The Monthly is the check of the system: how the company is trending and which learnings are worth keeping. A person accepts or discards each one.
The Quarterly is typically a workshop in which the quarter's figures and the next quarter's goals take shape, much as many companies hold an OKR workshop of a day or longer. If you plan in trimesters, you use a trimester meeting. If your company works with OKRs or EOS, the artifacts of those methods can be created and checked here. The yearly meeting is a strategy workshop: the VP Strategy works with the strategy advisers, and you choose the method, for example Blue Ocean. The Navigator leads question by question through purpose, values, vision and strategy, and then you set the yearly goals with measurable outcomes.
What is agreed in a meeting moves on. When the meeting closes, the goals go to the AI heads of the areas. Each head drafts the plan for its area, and the draft waits in your Inbox for your approval. Every goal is linked to its area and to a parent, yearly goals to the vision and all others to an outcome. The product refuses a goal without a parent.
Numbers, views and decisions
Every outcome gets a metric with a target value and warning lines. A traffic light shows the status and a responsible person looks after it, and you do not have to type the value: it can also be read automatically from a connected system. At the top of Rocket Routine you switch between Dashboard, Control Tower and Chat. The Dashboard is your personal overview. The Control Tower is the work board: work items from Backlog to Released, with exceptions such as blocked or escalated in a column of their own. Alignment shows how goals and outcomes connect. And you simply ask "How are we doing?": the VP in charge answers from the current numbers, on request with a chart.
Every decision lands in the decision log with an impact and a method. The impact says how far a decision reaches, in four levels. The Root concerns business model, financing or ownership and belongs to the CEO and the owners. The Trunk changes how the company is built and belongs to the leadership level. The Branch governs how an area runs and belongs to its responsible person. The Leaf is a routine decision with clear limits that an AI agent makes itself within its level. The method records how you decided, with six to choose from: with one, a responsible person decides alone when things must move fast, with another, the group reaches consensus. The Navigator leads you through the method, and one click on "Record this decision?" enters it.
You set how much an AI does on its own
There are three adoption levels. At Shadow an AI only makes suggestions: a form opens, and nothing is saved until you confirm. At Copilot it does everyday things itself, such as setting a status or entering a metric value, and for structural things like a new goal it opens the filled-in form. At Autopilot it does structural things itself and tells you in the chat. A person decides when an AI moves up a level, and the evidence is how often its work is right the first time.
An AI acts with exactly the rights of the person who asked, and at every level it asks before it deletes anything or changes the company's direction. Nothing leaves the company without your yes: a post or an email waits as a draft until you approve it. Work that needs a check is checked by a person against a definition of done, a checklist of what "finished" means, and never by whoever did it.
Proven methods under the hood
Rocket Routine OS works with methods that operations teams trust, and does the work of applying them. PDCA is the loop the whole system runs: Plan is goals, outcomes and work items, Do the work and the routines, Check the metrics and quality checks, Act what the team learned. Poka-Yoke prevents mistakes instead of finding them later: incomplete entries are refused, and numbers are read back before they are saved. The 7 types of waste are carried over to knowledge work: every quarter the VP Quality & Excellence looks for lost effort in your real figures and proposes what to remove. You do not need to know the methods to benefit from them.
Your day, your night
Routines and scheduled jobs run at night or on any schedule. You can simply close your laptop: the agents keep working. An intake routine reads your company's incoming channels, such as email, at the time you set, ignores newsletters and noise and drafts work items from real requests. Everything that comes in is triaged, meaning sorted and distributed: to your Inbox, to an AI agent, or to a colleague. When a routine finds something that needs a conversation, the avatar of the VP in charge pulses in the sidebar, and one click opens the chat. When you arrive at work, drafts, reports and finished work are already sorted.
A year in short
On day one you start a chat with the Navigator. It guides you through set-up in about 20 to 30 minutes, and you can pause at any time. You can start the set-up conversation right away. During it you connect your own AI subscription, and from then on your own AI does the work. The Navigator reads your company from its website and proposes purpose, vision, mission and values, which you confirm. It invites colleagues. If you already have goals in another tool or a spreadsheet, it maps them to vision, yearly goals, outcomes and metrics, shows you a table and creates them only after your OK.
In the yearly meeting, strategy and yearly goals take shape. In the quarters the VPs set their quarter goals under the yearly outcomes, and work items appear in the Control Tower. At the end of each quarter the meetings score each outcome with evidence and derive the next goals. At year-end you have a year you can read back: versions, decisions and evidence stay readable.
Who it is for
Rocket Routine OS is for companies led by founders, of any size. Anyone starting alone has the AI heads as a leadership team. When the first colleague joins, you give her an area, and the AI head supports her.
Rocket Routine OS is under active construction and comes in 2026. There is a waitlist and there are pilot customers, and we are not naming a start date yet. If you want to follow progress, join the waitlist.
Company 0
Rocket Routine GmbH is the first company that runs on Rocket Routine OS. Sven O. Rimmelspacher calls it Company 0. We do our own growth work with the product: an AI agent plans, drafts and delivers, and Sven O. Rimmelspacher leads and approves. This article came about that way too.
Frequently Asked Questions
What is Rocket Routine OS?
An agentic operating system for your company with an AI leadership team built in. Yearly plan, goals, numbers, meetings and daily work come together in one place. You talk to the Navigator and the AI heads of the ten areas, and they answer and do the work.
What is an AI head?
An AI head is the AI agent of an area and its VP, meaning its lead. It is named VP plus area, for example VP Marketing, knows the area's goals, metrics and work, and answers "How are we doing?" with exact numbers. A company can also make a person the VP.
What can the agents do?
Price lists, complete landing pages, social-media automation, projects and software, including large systems. They produce artifacts such as reports, presentations and web pages and use the browser. Any tool can be connected, and today there are about two dozen connections.
Does the AI act without me?
You set that for each AI on its own: Shadow suggests, Copilot does everyday things, Autopilot also does structural things. At every level it asks before it deletes anything or changes the company's direction, and nothing leaves the company without your yes.
Who is Rocket Routine OS for?
For companies led by founders, of any size. The same system fits one person, a small team or a larger team. Currently there is a waitlist and there are pilot customers.



