What we do
We build and run local AI infrastructure for companies in Munich. You keep using the tools your team already knows; the model moves into your building.
Four things
- Model and hardware selection. We match the model class your work actually needs to the smallest machine that serves it at your peak load, and we say plainly when the honest answer is to stay on cloud.
- Setup. Hardware procurement, installation in your office or rack, inference server, and an endpoint your existing tools can point at.
- Maintenance. Updates, monitoring, model upgrades as better open weights are released, and a person to call when something stops.
- Hybrid. Local for the bulk of the load, cloud for the few requests that need a frontier model. Most teams end up here rather than at either extreme.
How an engagement starts
- You run the cost analysis on this site. It takes a few minutes and gives you a break-even month and a sized setup.
- We go through the result together and check the assumptions against your real workload, not the defaults.
- If the numbers hold, we quote the hardware and the setup. If they do not, we say so and you have lost nothing but an afternoon.
On pricing
- Hardware is quoted at cost plus a setup fee that depends on the size of the installation. Maintenance is a monthly retainer. We publish the ranges once we have enough installations behind us to quote them honestly.
Why Munich
- Local hardware needs someone who can stand next to it. We work with companies we can reach in person.
Who you would work with
Two co-founders. You deal with them directly, from the first call to maintenance.
Or email regu.initial@gmail.com directly.