At some point deep learning will probably get to the point where you point a cell phone camera at the doodad, then prompt "wrench for this" or "missing piece", etc.