Deimann OS
The internal operating system behind every Deimann Com project — AI-native from day one.
Fact sheet
Operating a portfolio usually means hiring an organization.
Running parallel products — content, marketing, customer communication, internal coordination — traditionally means scaling headcount with every new project. That model breaks for a bootstrapped operator and gets more expensive with each product added to the portfolio.
An AI-native operating layer.
Deimann OS is the shared infrastructure behind every project in the portfolio. AI agents handle the recurring layer of content production, customer touchpoints, monitoring and internal reporting, while the founder sets direction and reviews the edge cases. Every product added to the portfolio reuses that layer, so the cost of the next one keeps falling.
Where it came from
Deimann OS wasn't planned. It accumulated. When you run several small products alone, you keep solving the same problems: something needs an image, something needs to know if it's still online, something has to store a lead, something has to remember what you decided last month. The first time you solve each of these it's a script. The third time you notice that the script belongs to none of the products and all of them, and you move it into a shared layer. After a year of that, the shared layer is bigger than any single product, and it has a name.
How it runs
Today the OS is a set of internal services and scheduled agents that every property in the portfolio uses. There's a watchdog that discovers each service on its own, checks a health contract, restarts what fails and reports to Slack, which is why most incidents end before I've noticed them. There's one image service that produces product imagery, essay covers and card motifs for every site from a single prompt system. There's a central lead inbox, a shared suppression list, hubs that pull search and behaviour data for all properties, and weekly loops that analyse them and turn findings into work. And there's the part that's hardest to explain: a machine-readable memory of decisions, edge cases and hard-won context that every AI session reads before it starts. It's the closest thing the company has to institutional knowledge, and there's no institution.
What I learned
The compounding is real but it's not free. Every service you build once lowers the cost of the next product, which is the whole thesis, but it also means every product now shares a failure mode. A bad day for the image service is a bad day for every site at once. I've had that day. The other lesson is that an operating layer is only as good as its memory. Agents that run on a schedule are tireless, not smart, and they only get smarter if what they did last week changes what they do this week. Most of the recent OS work has gone into exactly that: ledgers, policies, review dates. I wrote about it in Loops Don't Learn.
Status
Internal. Not offered as a product, not screenshotted, not for sale. It runs on European infrastructure, and the tools it's built on are listed on the stack page. If you want to understand how one person operates a portfolio, this is the thing to understand.
More projects
Deimann Com is a privately held software company from Hamburg that builds, acquires, and operates digital products on autonomous AI systems.
