We build digital products – and the companies around them.

Small, and deliberately so. There is a permanent core plus a network of specialists who join depending on the project. For you that means you speak to the people who actually build your project. For very large programmes with several parallel teams we are not the right partner – and we will say that in the first conversation rather than let you find out later.

From two directions. One is building itself: websites, apps and our own products, which we run to this day – from the Flutter app in the stores and the SwiftUI app across Apple platforms to websites that have to hold their rankings. The other is the client side: project management and process work on IT, ERP and e-commerce programmes, plus replacing overgrown Excel and VBA systems. That is where you learn what projects actually fail on – rarely the technology, usually unclear requirements and unclear ownership. Together those two became the double role vincubate runs on today: building and steering with the same people.

Our base is Dresden, Germany. Across the German-speaking region we are happy to come to you for workshops, kick-offs and working sessions. We do not have an office outside Dresden. Most of the work happens remotely anyway, including with clients outside Europe – see Markets & remote work.

Neither – we are a company builder. We develop our own products and ventures, and alongside that we take client projects that match what we are good at. In practice you notice the difference because we think like a product team: what does this thing have to achieve, what does it cost to run, and what happens to it in two years?

The core of the work stays with us, and you talk to the people building it. For bounded topics – design, native development, specialist data work – we bring in fixed partners from our network. We say so up front, with name and role, not on the invoice. We remain your contractual counterpart; responsibility does not travel onwards. If your policy restricts the use of third parties, we follow it or we decline.

No, and we are explicit about it: those are our own projects. Pauly, Chess Clock, Mini Marvin and the platforms in the portfolio were built and are operated by us. Client work is only shown with permission – the Webflow reference on this page is one of those. Further examples with numbers are available on request in conversation.

Confidentiality: gladly, mutual, and before the first detailed conversation if you need it. References: the public client reference on this page is a Webflow site in real operation; further projects we name in conversation, as far as those clients have agreed. What we do not have is a public Astro client reference. We would rather tell you that than invent one – the rest of our delivery experience sits in the products we built and run ourselves.

Yes, throughout, and openly. AI noticeably speeds up implementation, research and testing. What it does not replace is deciding what should be built at all and where the traps are. We also ship AI as a product – for example as a chat agent, or as a way to maintain a website by chat.

Four kinds, regularly. Plans that need considerably less than they are about to cost – then we describe the smaller version. Programmes with several parallel teams and staff functions, for which we are too small. Pure body leasing over months with no outcome of our own. And anything where, after the conversation, we do not believe that speed, budget and scope can all be true at once. Refusals come early, not after four weeks of proposal prose.

Completely. Code lives in your repository, hosting and stores run on your accounts, credentials are handed over. If you want to continue with someone else, it costs you nothing but an email. We consider that the only honest basis for a long working relationship.

It depends on scope – and we know the scope after the first conversation, not before it. That is why there is no fantasy price list on this website. After the conversation you get a traceable estimate along with the assumptions it rests on, so you can see what moves the number. The MVP validation sprint is fixed price.

Both, depending on how sharp the scope is. When it is sharp we work to a fixed price – that is how the MVP validation sprint runs. When it is not, we bill for time spent against an agreed cap, invoiced monthly, so you can stop at any point. Scope changes are named as such and quoted separately instead of quietly eating the budget. What drives the number on websites is written out under Astro cost.

First: you hear it from us, not from the calendar. Delays and misjudgements get reported the moment they become visible, with a proposal attached rather than just an alert. Anything we built that does not work as agreed, we fix under statutory warranty and without a new invoice. And if the collaboration still does not hold, it ends without drama: code, repository and credentials are yours anyway, the state is documented, and another team can pick it up.

Usually a project or framework contract with a statement of work, a mutual non-disclosure agreement and, as soon as personal data is involved, a data processing agreement. We review your purchasing terms and tell you what we can carry and what we cannot. On insurance we deliberately publish no coverage figures here for you to take on trust: which certificates your procurement needs, we clarify before signing and then provide.

We ask for the narrowest access that works, on our own accounts with two-factor authentication, and it is returned or revoked when the project ends. We only take production data when the task genuinely requires it; otherwise we work with anonymised or invented test data. Credentials and personal customer data do not go into AI tooling – that is a rule here, not a judgement call. Every service provider involved gets named so your data protection officer can put the agreements in place. More on that under GDPR-compliant websites.

German and English, both at working project level – requirements, documentation, code comments and meetings. This entire website exists in both.

Yes, after an audit. We look at the code, the structure and the dependencies and tell you honestly whether taking it over makes sense or whether a rebuild ends up cheaper. Sometimes we advise against it.

Continuously. There is no application deadline – we review as things arrive. Open roles and the co-founder models are in the careers section (German).

Thirty to forty-five minutes, video or phone. You describe the plan; we ask about the goal, the constraints, the timing and what has already been tried. At the end we say whether we fit. Free, no obligation, and no follow-up meeting if it does not.

Usually within 24 hours on working days – and with a substantive answer rather than a receipt confirmation. When we can actually start depends on the projects already running; we name a date for that instead of keeping you waiting.