Hey, thanks for stopping by.

I wondered what it would take to control my own AI stack and prevent my professional work from being locked into a few big companies, so I started Contenox.

I’m Alexander Ertli, founder and maintainer of Contenox.

Contenox started as a broader LLM proxy and workflow engine for autonomous AI. In November 2025, I narrowed it toward the problem I actually care about: AI for engineers that runs locally, has explicit workflow boundaries, and is useful for day-to-day work on hardware you may already own.

The point is not another black box that pretends to be autonomous. The point is a system you can inspect, run, repeat, and change. Work should be packaged into reviewable chains, not hidden behind a chat session you cannot reproduce.

That is also why the runtime work goes below the prompt layer. Contenox uses modeld, my shim around local inference backends such as OpenVINO and llama.cpp. It inverts the usual vLLM-style server-first view: the workflow and its state are the center, and model execution is a local component that should fit into that boundary.

My professional work still sits around cloud and application architecture, SaaS, ERPNext, Kubernetes, Terraform, GitOps, and production systems. Contenox is where those concerns meet my own AI stack: local-first, explicit, boring enough to operate, and useful enough to keep using.

• Contenox
• CV
• Get in touch