Harry Potter had a farm
As strange as it sounds, the most successful British products of our age are “The Lord of the Rings” from a time when the British Empire still ruled the waves, and “Harry Potter” from a time when the British Isles hardly mattered at all, both turned into blockbuster movie franchises by Hollywood. In a sense, that involved the downgrading of the beautiful and ultra-proud High Elves of Middle-earth into the ridiculous (though highly capable) House Elves.
A few decades before Tolkien penned “The Hobbit,” 19th-century German and American companies began smothering the obsolescent British manufacturing. Even the continued British advances in natural sciences and engineering design could not change the grim reality on the shop floor. To make matters worse, the contemporary British elite of the 1880s committed themselves to free trade paradigms, refusing to be more aggressive against their international competition.
Over the last decades, while the collective West has found itself invading others, being invaded, and experiencing both terminal spiritual decline and deindustrialization, China was welcoming Western manufacturing fleeing from rising costs and suicidal green and labor policies, and harvesting their know-how. Today’s China is an integrated and almost entirely sovereign industrial juggernaut capable of beating the best American companies at their own game.
The second Trump administration will be focused and ruthless. They will eliminate stifling regulations, expand access to natural resources, and open up a huge mercantilist umbrella targeting China. Yet, it will be entirely up to American enterprises to use that opening and swiftly optimize their processes.
Despite the upcoming tsunami of industrial automation, a significant chunk of American manufacturing will continue to employ large numbers of humans, relying heavily on their access to product assembly and maintenance manuals. These manuals must be created and updated in the most accurate and timely manner. Presently, they are anything but.
The situation is particularly serious for large companies that do not own all the data in their products. An aerospace OEM buys a landing gear, which consists of a few large assembly components from suppliers, and installs it. Later on, during maintenance, parts in those assemblies need to be replaced. Data about those parts and their relationship to the next higher-level assembly is not stored in the OEM’s PLM; it is most likely documented exclusively in a PDF manual provided by the supplier.
Today’s technical publications for aerospace and defense derive mostly from CAD models. They can either be PDFs created according to older FAA ATA chapters or a combination of a far more advanced S1000D-based package with a PDF generated from that package. In either case, manufacturing personnel often resemble House Elves, printing these manuals and tediously sorting them out on the shop floor. Unfortunately, since the connection between the original 3D model-based sources of truth and the markup-based publications is severed at this point, those House Elves struggle to detect inevitable engineering changes and incorporate them back into their processes, which is critical for maintenance operations.
The Senticore team is highly sympathetic to the House Elves, and we have been working for quite some time to address their grievances. Here is our thinking on the correct path forward:
- The ability to quickly and reliably find relevant parts or assemblies associated with a particular manufacturing or maintenance operation requires fast access to the associated 3D data, which traditionally resides in its secure PLM/CAD silo
- Furthermore, part number and revision alone lack the level of uniqueness that House Elves need in their context. A more intelligent unique identifier that includes spatial information for that specific instance might be required.
- Fast access can only come from good spatial indexing, and so far, no major PLM/CAD vendor has provided a solution to handle the data load characteristic of aerospace’s large models. The only fast and scalable spatial indexing solution we have seen comes from our friends at Spacis.
- While spatial indexing is necessary, it is not sufficient. Sufficiency comes from the ability to explicitly link the unique IDs in PLM/CAD data to specific nodes in a publications data model. There is doubt that such linking and updates can be completely automated; perhaps we are speaking more about a smart notification engine and human-assisted validation.
- Speaking of nodes, while reconstructing a comprehensive and conveniently granular data model from an S1000D package is relatively straightforward, doing the same for PDFs is much more complicated because these publications are very complex infographic documents. We believe we can deal with it with a daisy chain of several distinct AI technologies.
There are a multitude of visually beautiful and conceptually interesting scenes in “Harry Potter,” yet I consider the moment when Dobby receives the sock and then reveals his true potential as the most striking. Apparently, both the good and bad guys had access to this simpletonish yet powerful resource all along, and they all failed to fully utilize it to their advantage.
The great powers of today are riding the waves of space tech, drones, and AI to position themselves for the next confrontation. Engineering and manufacturing efficiency are crucial in that struggle, and solving the shop floor’s documentation conundrum might be one of the untapped pieces that can truly sway the balance for American manufacturing at this critical moment.
Talk to us, and let’s give your House Elves the socks they deserve!
