// OUR STACK
Open source, top to bottom.
Every tool we use to run this business is open source, self-hosted, and running on our own hardware in the EU. We recommend only what we run ourselves.
€0
Monthly SaaS cost
Never
Data leaves EU
None
Vendor lock-in
100%
Open source
// AI / LLM BLUE
vLLM/LlamaCpp Local LLM runtime. Runs Llama, Mistral, Qwen, DeepSeek on our own hardware. OpenAI-compatible API, zero external calls.
Qdrant Self-hosted vector database powering all RAG applications and document intelligence demos.
ComfyUI Local image and video generation pipeline. Brand-consistent output at zero marginal cost per run.
// Automation AMBER
// Infrastructure TEAL
Coolify One-click Docker deployments, automatic SSL, built-in backups. Every service managed from one interface.
Traefik Handles all traffic routing and SSL termination. Every new service gets HTTPS automatically.
Forgejo All source code, client repos, workflow exports. No dependency on external code hosting.
Grafana + Prometheus Cluster health, LLM inference latency, workflow stats. We know when something is wrong before clients do.
// Website & content PURPLE
Ghost Blog engine and newsletter platform. Webhooks trigger the full content pipeline on publish.
Astro This website. Pure HTML output, perfect performance scores, zero server runtime.
Rybbit GDPR-compliant analytics. No cookies, no consent banner, no data leaving our infrastructure.
Postiz OSS social scheduling. Post published — LinkedIn, Instagram, and Mastodon updated automatically.
Local AI cluster
Mini PC cluster handling local LLM inference, generative workloads, and development. The same hardware we recommend to clients starting out.
Full build guide →Coming in a blog post
EU production server
Dedicated VPS on European infrastructure, physically located in France.
Ubuntu 24.04, containerised via Docker, automatic SSL. All production services run here.