01 / AI
AI Process Automation
Agents, RAG pipelines and LLM integrations for real business processes — from document workflows to structured back-office automation.
AI Process Automation × Infrastructure
AI-driven business processes — on infrastructure you control.
Cloud, on-prem, or hybrid. We build the automation — and the foundation that runs it reliably.
Four pillars, one accountability.
AI apps rarely fail at the model. They fail at data, deployment and operations. We deliver both, from one team.
02 / Cloud
Cloud Architecture
AWS, Hetzner, hybrid: landing zones, IaC, security baselines, FinOps. So scale doesn't become an accelerant for chaos.
03 / On-Prem
On-Prem Stack
Kubernetes (k3s), Vault, local LLM inference with Ollama. When data must stay on site, we run it where it lives.
04 / Operations
Migration & Operations
GitOps, observability, on-call runbooks. We don't just hand over a stack — we leave a team that can actually run it.
Why us
AI without infrastructure is a demo. Infrastructure without AI is just admin.
Most AI initiatives ship impressive prototypes — and stall at the leap to production. The model runs, but no one knows who deploys, monitors, versions, or hardens it for compliance.
We connect both sides: we build the AI workflow and the pipeline that makes it reproducible. Cloud when it makes sense. On-prem when it has to. Hybrid when reality is in between.
A model is only in production when someone, at 3 a.m., knows what to do without picking up the phone.
Three phases — no waterfall.
Start small, prove value in weeks not months, scale only what worked.
- 01
Discovery
Map processes, identify ROI candidates, assess infra maturity. Output: a prioritized roadmap with explicit hypotheses.
1–2 weeks - 02
Pilot
One use case end-to-end: data integration, model, deployment, monitoring. On your infrastructure, with your team.
4–6 weeks - 03
Scale
More use cases, more automation, fewer manual handoffs. Plus knowledge transfer so you can keep building.
ongoing
Tools we run in production.
No buzzwords — just what we use every day.
KubernetesArgoCDTerraformOllamaLangChainPostgreSQLAWSHetznerVaultPrometheusGrafanaGitHub Actions
What we've built.
A snapshot of recent work. Details on request.
Toys & Media
Audio transcription with speaker diarization for content pipelines
Kubernetes-based service for multi-language audio transcription — including speaker recognition, on-prem inference, and cost monitoring.
→ 5× faster processing
Construction
Cloud migration with Terraform IaC and compliance-aware tagging
Full AWS landing zone with security baseline, cost allocation, GitOps deployments, and connectivity between legacy and cloud workloads.
→ 100% IaC coverage
SMB / Consulting
AI platform with local LLM inference over WireGuard
k3s cluster on Hetzner paired with a Mac-Studio-hosted Ollama for Metal GPU inference. ArgoCD GitOps, Langfuse observability, fully automated deploys.
→ <€30/month infra
Let's build something useful.
30 minutes, no sales pitch. We listen to what you're wrestling with — and tell you honestly whether we can help.