Four ways out of the day-to-day.
Each one states who it is for, what happens in the first week, what belongs to you at the end, and what is explicitly not included. That last point saves us both the wrong conversation.
Operations that run without you
Businesses where the same manual work comes up every week: transferring orders, chasing quotes, typing data from here to there.
I look at what actually happens, not at what the process document says. By the end of the first week you know which task is worth automating and which is not. Sometimes the answer is that the process upstream is broken and automating it would only break it faster.
The processes run under your own accounts, documented in your language. If you let me go, everything keeps running.
Not included: No ongoing retainer you cannot get out of. And no platform that is yours only for as long as you keep paying.
Systems that speak up
Anyone who has already automated things and honestly does not know whether they still run.
I check whether your existing processes actually do anything or merely look green. That is a difference more often than people expect: in my own setup it was three out of a dozen.
Watchdogs that can tell "measured and bad" apart from "not measured at all". Plus an alert that tells you what to do, instead of just that something is red.
Not included: No dashboard you have to look at every day. A system that speaks up on its own beats one that needs supervising.
Ready-made blocks
Anyone who wants to install it themselves and just needs a proven template.
There is none. You buy, download and install it yourself. Every pack ships with a guide that starts at zero.
The code is yours. No subscription, no licence check, no dependency on me.
Not included: No installation, no support contract. If you want it installed for you, it is not a pack, it is a project.
See the blocks →Workshops, 1:1 or small group
Founders who do not want it taken off their hands, they want to be able to do it.
We work on your real process, not on a textbook example. You leave with something that runs at your place.
Possible modules: apply the autonomy tiers to your own business, take one real task all the way to a running process, make sure the AI gets the right information without you assembling it every time.
Not included: Nothing is bookable yet. Mention it on the intro call if the format interests you and you get it first.
The same Tuesday, twice.
On the left, what a day looks like when everything goes through your hands. On the right, the same day once the processes are in place. Flip it.
- 01An order arrives by email, you type it into the systemThe order is already in the system, you just see a line for your information
- 02You check three times a day whether something is stuckThe system tells you when something is stuck, and stays quiet otherwise
- 03The customer enquiry waits until you have time in the eveningThe reply is sitting there drafted, you read it and send it
- 04At the end of the month you roughly know how things wentYou know at any time, because the numbers travel there by themselves
- 05The one task only you can do sits untouched when you are illThe task is written down and keeps running, including without you
Four steps, not a waterfall.
Intro call
25 minutes. You describe one task, I tell you honestly whether it is worth it.
Look closely
Before anything gets built, I measure what actually happens. This is the step almost everyone skips, and the reason so many automations start at the wrong end.
Build
Fixed scope, fixed end. You see the intermediate results and can correct them before the next step builds on top.
Hand over
Accounts, documentation and a guide someone else could carry on with. I am replaceable too, and that is exactly the point.
Let's talk about one process for 25 minutes.
Free, no strings, by video or in person in Berlin. You get an honest assessment at the end, including if that assessment is that you do not need anyone for this.
My commitment: if a handed-over process does not run the way we described it beforehand, I rebuild it until it does. No new invoice, and no argument about whose fault it was.
Pick a slot.
Choose a time that suits you. The few questions in the form are there so I arrive prepared.
Clicking loads the Cal.com calendar, which sends data to Cal.com. Nothing is loaded before you click. Privacy
Six questions that always come up.
Answered honestly, including where the answer is inconvenient.
I already build most of this myself with Claude. What do you do then?
The part you skip because it is no fun. You build the thing that works. I build what keeps it working: error handling, retries, credentials, monitoring, and a structure where the knowledge does not live in your head and a chat log. That is roughly sixty percent of the work and zero percent of the fun. If what you already have runs without you, I will say so on the intro call and we are done in ten minutes.
What does it cost?
There is no price list, because no two businesses are the same. After the free intro call you get a fixed price for a clearly defined scope, not an open-ended hourly rate. The call itself costs nothing and commits you to nothing. If it turns out that automating this does not pay off for you, I will say so, and then it was a free conversation rather than an expensive project. And if a handed-over process does not run as described, I rebuild it at no extra charge.
How long does a project take?
A clearly defined process is usually running in one to three weeks. The first step is never building, it is measuring: what actually happens in your business, as opposed to what the process document says. That step costs a few days and decides whether the rest is worth doing at all. Larger plans get split into individual processes that each stand on their own, rather than months of work towards one big whole.
What happens if I stop halfway through?
You keep everything that is finished, and it keeps running. Every process runs under your own accounts, every account is in your name, the documentation sits with you. There is no platform that is yours only while you keep paying, and no licence I can switch off. That is not goodwill, it is the reason I build this way: a system you cannot leave is a bad system.
What happens to my customer data?
Processing stays in the EU: servers in Germany, services with a European data centre, and a data processing agreement for the engagement itself. But the cheaper question comes first: what data does a process actually need. A field that never travels is a field nobody has to protect. Anything sensitive, health records, job applications, payment details, is never shown to an AI model; fixed rules handle it instead.
How much is the AI allowed to decide on its own?
As little as possible and as much as necessary, and that is settled up front rather than discovered in production. Every task gets one of four tiers, sorted by four questions: Can it be undone? Who does it hit when it goes wrong? What does one failed attempt cost? Does it touch brand, money or law? Anything that reaches a customer stays at the most cautious tier until there are good reasons for something else.