Open source is infrastructure⁠.

Open source is not a licensing detail at the edge of my work. It is the foundation of much of the technology we use every day. Operating systems, databases, frameworks, models and developer tools only exist in their current form because people share their work and allow others to build on it.

Openness changes what is possible

Projects such as Mistral and DeepSeek show that capable models do not have to exist exclusively behind the APIs of a few providers. I am less interested in which model happens to win a benchmark. What matters is whether weights, code or technical approaches become accessible enough to inspect, adapt and use in different ways.

The same applies to tools such as Ornith. An open coding agent is not merely a finished product I can use. I can understand how it works, examine its limits and build my own ideas on top of it. That transparency matters even more as systems increasingly generate code and structures themselves.

Local models add another layer of freedom. They are not the best answer to every task. But they provide a real choice: Where does my data run? Which dependencies am I willing to accept? What can I operate myself?

Independence takes work

Open does not automatically mean secure, simple or free. Software still needs maintenance, documentation, testing and updates. Independence therefore requires more than access to source code. It also requires knowledge and a willingness to take responsibility.

That possibility still matters. Especially in Europe, we should be able not only to consume AI, but also to understand and operate it ourselves. This is not about rejecting large providers. I am happy to use their systems when they fit the task.

I simply want them to remain an option rather than becoming the only available infrastructure. Good technology should be open to inspection, adaptation and further development. That is why open source is not a niche to me. It is infrastructure.