Proprietary Semantic Layers And Proprietary I/O Cards For Personal Computers

EDITOR’S NOTE: This post was inspired by today’s opening keynote for the Semantic Layer Summit 2025, Presented By ATSCALE.

In my previous post, I wrote about the following running poll in my favorite LinkedIn Group – Purchasing & Global Supply Chain Professionals. Again, please cast your vote and share your thoughts.

The poll’s question is pretty straightforward: Without industry experience and expertise, you may not know the right questions to ask GenerativeAI, let alone understand the answers.

It is a fair question and one about which I have previously written in the following post: Tell me how Agentic AI would have known to ask, “What time of day do orders come in?”

With Generative AI or Agentic AI, you are the “true knowledge source” and therefore the key to empowering AI, rather than AI empowering you. I would like you to consider the following question I posed to a Generative AI platform regarding the similarities and differences between proprietary semantic layers and personal computer I/O boards.

QUESTION: What are the similarities and differences between proprietary semantic layers, proprietary I/O cards for personal computer motherboards, and when and why did they do away with the latter?

This is a sharp and insightful analogy. Comparing proprietary semantic layers in modern enterprise software to proprietary I/O (Input/Output) cards in early personal computing highlights deep economic, technical, and strategic parallels. Here’s a breakdown of their similarities, differences, and historical lessons:

---

🔄 SIMILARITIES: Semantic Layers vs. I/O Cards

Aspect	Proprietary Semantic Layers (Today)	Proprietary I/O Cards (1980s–1990s PCs)
Vendor Lock-In	Semantic layers tie users to a platform’s data logic	Custom I/O cards locked users into specific hardware or OS environments
Incompatibility Issues	Can’t easily integrate across platforms or ecosystems	Hardware couldn’t be reused or upgraded without matching drivers or slots
Fragmentation of Standards	Each vendor creates its own logic layer (e.g., Coupa vs. SAP Ariba)	Each vendor used unique pinouts, drivers, and BIOS-level protocols
Barriers to Innovation	Hard to adopt AI across tools with different semantics	Hard to scale, upgrade, or interconnect devices without standard interfaces
Complex User Burden	Users must learn vendor-specific data logic	Users had to configure jumpers, IRQs, or BIOS settings manually

In both cases, closed ecosystems benefited the vendor but hindered user efficiency, integration, and cost-effectiveness.

---

🧬 DIFFERENCES

Aspect	Semantic Layers (Software)	I/O Cards (Hardware)
Domain	Software & data abstraction (semantics, ontologies, logic)	Physical hardware interfacing (printers, modems, sound cards)
Customizability	Layers can be abstracted or mediated by knowledge graphs	Hardware was less abstract; physical form and electrical specs mattered
Decay of Relevance	Semantic layers are still evolving, tightly integrated with AI	Proprietary I/O cards were eliminated as universal interfaces emerged
Persistence	Semantic lock-in persists through data models and AI pipelines	Hardware lock-in faded with the rise of USB, PCI, and plug-and-play tech

---

🕰️ WHY AND WHEN THEY DID AWAY WITH PROPRIETARY I/O CARDS

What happened:

• In the mid to late 1990s, proprietary I/O cards started to disappear as standardized, open interfaces took over:
  • PCI (1992) and PCIe (2003) standardized expansion slots
  • USB (1996 onward) unified device interfaces and eliminated custom drivers
  • Plug and Play in Windows 95 enabled automatic hardware recognition
  • Open driver models like UAA (Universal Audio Architecture) reduced need for vendor-specific software

Why it happened:

Reason	Impact
User demand for interoperability	Business and consumer users pushed for easier integration
Hardware commoditization	Hardware became cheap, performance was no longer the differentiator
OEM pressure	PC makers pushed back on vendors to reduce costs and complexity
Standard governance bodies	Organizations like IEEE, USB-IF, and PCI-SIG pushed for open standards
Developer community pushback	Developers preferred APIs and SDKs that worked across hardware

---

📌 LESSON FOR TODAY’S SEMANTIC LAYERS

Lesson from Hardware History	Modern Semantic Layer Implication
Proprietary standards give way to open protocols	Expect future demand for open ontologies and knowledge graphs
Compatibility wins over lock-in in the long run	Interoperability will determine AI adoption and value realization
Vendors who embrace open ecosystems gain trust	Solutions like AdaptOne, Denodo, Stardog, etc., offer abstraction layers for semantic harmonization
User-centric design drives evolution	Data stewards and procurement teams will demand explainability and portability

---

🏁 Bottom Line

Proprietary I/O cards faded as users, developers, and OEMs demanded freedom and interoperability.
Today’s proprietary semantic layers will face the same pressure—especially as AI, analytics, and data collaboration demand shared understanding across tools.

30

The Rise Of The Proprietary Semantic Layer

The Projected Fall Of The Proprietary Semantic Layer

What impact does the rise and fall of proprietary semantic layers have on ProcureTech, procurement, and supply chain?

The rise and projected fall of proprietary semantic layers will have transformational consequences for ProcureTech, procurement, and supply chain operations, influencing everything from platform interoperability and supplier collaboration to AI reliability and cost control.

I will expand on the above impact in my next post. That said, the takeaway from today is the need for critical thinking – your critical thinking, to make Generative AI and Agentic AI effective. In short, you make AI smart; AI alone doesn’t make you smart.

30