IoT Devices as Layers: But What Is There to Autocomplete?

Connected consumer devices form an installed substrate of electricity, connectivity and heterogeneous computing resources. Their owners may control the physical equipment while suppliers retain authority to update its software and direct its behavior through device-initiated management sessions. This paper examines a conditional AI-wormnet scenario: a self-propagating distributed workforce that sustains learned inference without continuously depending on a centralized model service. Its workers need not contain identical language models. They could combine small decision models, task specialists, conventional software and compatible shards of larger networks, assigning roles according to available hardware. We propose that the relevant unit of persistence is a task-bearing configuration, rather than a copy of one model. A four-plane framework separates compute, task continuity, connectivity and consequential authority, with provisioning trust as an independent dimension. An offline simulator illustrates how state loss, shared connectivity and action revocation produce different containment outcomes. The paper contributes an operational definition, a capability-based evaluation protocol and a reproducible structural artifact. It does not demonstrate autonomous infection, model competence or infrastructure compromise.