Everything stays in the building

When I started controlling installations, the meter cabinet was a wooden panel with relays and a Raspberry Pi. One thing was fixed from day one, and it has never changed: the controls run in the building itself. Not in a data centre, not at a manufacturer, not in an app that first has to pass through a server in another country.

There are two reasons for that, and both are practical.

The first is reliability. A heating system that fails because an internet connection stutters or a cloud service is switched off is not a good heating system. Manufacturers discontinue devices, services get acquired or disappear — the installation in a building should notice none of it. The Homer EMS therefore works entirely locally: if the internet drops, everything simply keeps running.

The second is that energy data ought to be private. Whoever can see at the level of hours and appliances what a household or business consumes, can read from it when someone is home, when work is being done, when not. That kind of data does not belong as by-catch on a supplier's servers. With Homer it stays where it originates: in the building.

That choice carries through into the hardware. What began with off-the-shelf modules has grown into self-designed circuit boards that measure and switch in the meter cabinet, with their own firmware. The same rule applies there too, but for failure: if the controls drop away, every switch falls back to a state in which nothing can break. A system that autonomously runs a building must also be well-behaved in what it does when it briefly isn't running things.

This is a recognised R&D programme in daily production at a fruit farm in Amsterdam; at a dry cleaner in De Pijp it is currently being set up. It is not an off-the-shelf product you can order; it is in active development, and I deliberately take on a limited number of custom and pilot projects. That way every installation remains something I can personally stand behind — from the foundation to the firmware.