The edge only works if the site can run itself
NodeWeaver’s public position is a hyperconverged edge cloud: self-managing, continuously optimized, zero-touch compute where the data and the process already are. Mike Grosswald (Partnerships) brought that to the August call as the NodeWeaver Edge AI platform.
Industrial teams already know the failure mode. A model that looks fine in a central lab still needs someone on site to patch, watch, and recover every box. Zero-touch is the claim that the platform does that work.
Where this sits next to the virtual twin
The same hour opened with Dassault Systèmes on Virtual Companions inside the twin. Edge infrastructure is the layer under that: the companion and the plant data have to run close to the line, without a new operations team at each site.
Practitioners should leave with a sharper question than “do we have an edge box?”: who keeps it healthy after the pilot, and does the twin still mean the same thing at the site as it does in design?