Bootstrapping a software factory. It is deliberately below the radar for now, so there is nothing much to point at yet.
The years since Xepelin have mostly been client work, and that turns out to be the point — the same people are becoming its first customers. Building the thing out of work that already exists seemed like a better order than building it out of a pitch and hoping.
Alongside it, learning AI properly rather than absorbing it, which in practice means running models on hardware I own instead of renting somebody else's. There is a homelab here doing rather more than it needs to. Cameras and vision systems that have to keep working when nobody is watching them. A 3D printer. Drones and robotics, still an interest rather than a project.