I know what you changed last summer

Some people are well aware of the bad things they did last summer, yet others are desperate to recall the intoxicated madness they unleashed on Vegas yesterday. Then, perfectly sober engineers change designs and configurations, sending the freshly generated PDF manuals to the shop floor far away, where harried folks in greasy overalls roll the dice trying to pinpoint the differences from the previous version. In either case, they can all end up on the hook from the urban legend vigilantes, customers, authorities, or the bloody mix of the above.

Considering the aforementioned mess, the natural instinct says to stay sober and get rid of the PDF-based data transfer in the engineering and manufacturing domain in favor of a unified platform such as 3DExperience, or an exchange based on a standard machine-readable format. While the former is a matter of personal decision, the latter options require significant collaboration between multiple free agents, and may have unforeseen long-term consequences.

Speaking of long-term, the issue of exchanging data between unrelated parties today is similar to the same company dealing with its own engineering treasure trove in a year, or maybe a few decades from the moment the data was created. A dependence on any particular vendor to interpret the business-critical data stored inside their application is a strategic vulnerability. This is hardly an exaggeration: upgrade and migration bugs happen, infrastructure and security disasters strike without warning, and overt and covert vendor locks are real.

Our project is about reconstructing PDF technical manuals into a JSON, ready for ingestion into other systems like MES, or for immediate conversion into a ready-to-use human-readable work-plan. Initially, Senticore considered it an extremely valuable yet temporary solution till the outdated documentation and systems fade away. Strangely enough, we presently observe a rollback to the legacy frameworks across the board manifested by faltering of entire countries, breakdown of formerly stable alliances and supply chains, and even splintering of global internet into quarantined enclaves.

With all the excitement about AI, protein-based intelligence still provides unparalleled balance of cleverness, flexibility, and dexterity. Hence, a transmittal package formed around human-readable PDF/A in combination with AP242, and S1000D seems to be the winner for both data exchange of today, and long-term archival for the future. As a reminder, PDF/A is an ISO-standardized subset of the PDF format, specifically designed to ensure that the PDF content is self-contained and can be reliably reproduced and viewed exactly as it was created regardless of the external ecosystem changes,

As far as it goes, continue leaning on a PDF for technical publications for as long as necessary, while adding S1000D content and turning PDF/A features on whenever possible.

While getting PDF summaries using out-of-the-box generative AI is already a daily convenience, it lacks reliability required for critical work-planning and subsequent versioning. In contrast, our project makes sure to reconstruct different versions of the same technical manuals in a way that allows for easy comparison, and the planning steps rearranged accordingly. Imagine you are dealing with the manuals 1,000 pages long, or several such manuals simultaneously, and the ROI for such a solution becomes self-evident.

For all that serious talk – it’s summertime; the bees are buzzing, and for many people it is time to relax. So, drink responsibly, drive safely, and let Senticore take care of the technical publications gap in your digital thread. L’chaim!