Fast and Furious on a Digital Thread
It feels like the world has been spinning particularly fast over the last year, and the past few weeks have been especially entertaining. The war in Ukraine rages on, and the Middle East literally blows up daily. China is rising while the Western schism accelerates. Artificial intelligence is eating the world, and individuals, companies, and entire countries are racing to figure out their place at that table.
In that overarching cacophony, there are many business processes where humans consume complex unstructured data and produce certain output. This involves analysis, generation, comparison, transformation, references, and logical links, often in large volumes. We see Senticore’s value proposition in applying assorted artificial intelligence methods to significantly optimize these processes.
As concerns about digital sovereignty grow and quasi-religious considerations increasingly drive supply chain and technology transfer decisions, how does that affect enterprise applications infrastructure, including PLM? When decades of plummeting birth rates, declining Western educational standards, and work ethics come to roost, who is going to replace the retiring old hands on the shop floor? How can corporate engineering and manufacturing teams balance the artificial intelligence FOMO vs. FAFO?
The United States is presently tripling down on restoring its industrial mojo from the failed “end of history“ and “flat world“ dreams. The American reshoring looks like a variation of a Fast and Furious movie franchise, where the protagonists drive while inspecting, maintaining, and even redesigning their car at every opportunity, often without stopping. In that paradigm, there’s a need for a clear business vision, continuous tracking of all relevant BOMs and configurations, and extreme reliance on the team’s motivation and expertise. In turn, expertise is fragile; it requires team members to continuously sync their knowledge with each other and the documentation.
Speaking of mojo, there’s a good reason Israel is eliminating Iran’s key nuclear scientists: hard-earned competence and the precious personal networks are key for any serious endeavor. Yet, over the last 30 years, the United States has discarded and given away to China entire layers of engineering and manufacturing expertise. Paradoxically, the United States may need to out-China China, both by hoarding back know-how from friends and allies like Japan and Korea (and even adversaries where possible), and by building it from scratch.
The Fast and Furious crew would hate to become vendor-locked, especially by their real or perceived adversaries. They accept the necessity of the moment, while always striving for flexibility. The issue for today’s industrial enterprise isn’t whether the PLM system of their current choice is good, but whether they can integrate it freely or even escape from it further down the road as circumstances change.
Similar to the Fast and Furious crew’s self-reliance attitude, a modern enterprise would be unwise to outsource the responsibility for its digital engineering ecosystem future. They can, however, build a roadmap (we can help with that), and use third-party tools to track their systems’ score regarding long-term adaptability and escape routes, and adjust accordingly. That means, for example, avoiding use of too proprietary CAD features and continuously testing the design data for an industry standard compatibility like AP242 (STEP/JT), and maintaining a separate graph-based copy of the released data and important downstream artifacts. Although this is hardly a new idea, its implementation became an order of magnitude cheaper with the right setup of the already proverbial MCP and intelligent agents.
On a separate note, we recently discovered unexpected benefits of multi-modal LLMs for PLM systems’ migration purposes, especially when certain vendors have done their best to obfuscate their data models to obscenity. AI is clearly capable of quickly uncovering relationships between various tables and attributes to build a fully coherent view of their environment.
The last thing the Fast and Furious crew wants is to veer off-track due to the temporary absence or a tragic loss of a team member; they always want to be able to bring in newbies and train them to the relevant expertise quickly. Here, there’s a path for American engineering and manufacturing enterprises: using generative AI to combine system prompts created by company veterans with uploaded documentation, bringing the most relevant knowledge to juniors’ fingertips without annoying the senior members unnecessarily.
Also, locally deployed LLMs are the new black in terms of ultimate digital sovereignty; whatever you ask it, whatever you upload into it, is yours and yours only. Incidentally, NVIDIA will soon ship RTX 6000 with 96GB VRAM, addressing many hardware limitations we have been experiencing during our development process, and providing a simpler blueprint for our customers.
Whether you are a startup or an established corporate Fast and Furious team in the engineering and manufacturing domain, talk to Senticore about your digital engineering practices. Drive fast, maintain the balance, jump over the pitfalls, and conquer every quarter-mile of the AI age. Remember: winning’s winning!
