Industry-Specific EAI Systems

This a timely topic for us at KYield. We developed an industry-specific executive guide in August and shared selectively.

In the first dozen we shared with Fortune 100 CEOs, it resulted in an immediate reply from one CEO and a forward from another tasking a senior member of his team to set up partnership discussions. Another triggered an internal referral. The guide also accelerated a collaboration with business owners and CEOs in a large industry on an industry-specific KOS.

So why is this topic suddenly so important? After all, we’ve been doing deep-dives with industry leaders for nearly 15 years. I think it’s a combination of factors, including but not limited to the following:

1. Security. Security is the top priority today whereas a decade ago the priority was typically deep, narrow use cases usually relating to increased revenue.
2. GenAI is necessary but insufficient . CEOs and boards are now better informed and increasingly realize they need more than generalized LLM chatbots.
3. Legal risks . Increased regulations (see articles 1 and 2 ), liability, and need for compliance are top of mind. ?
4. ROI , or lack thereof. After investing substantial sums for years in one-off ML projects, most of which are never commercialized, many companies are under pressure to show an ROI. The cost of our KOS has declined by 100x in the last decade, performance has increased even more, and it’s now accessible on devices via DANA (our digital assistant).
5. Sovereignty . Systemic cybersecurity risks , over-dependency on cloud providers, lock-in, and commoditization of AI favor entrenched monopolies (especially Big Techs). Everyone else must create, defend and enhance a competitive advantage.

Below is a clip from our executive guide on the industry-specific KOS, followed by the concluding chapter, “Evolve of Die”. (Note: LI editor wouldn't allow some links.)

Tailored to each business

KYield has performed deep-dives in most industries over the past 15 years. We are currently working on multiple industry-specific versions of the KOS in trillion-dollar + industries in collaboration with customers.

The exponential improvement in performance and cost over the past decade makes the KOS an attractive option for mid-market customers as well as industry leaders.

The universal KOS described in chapters 1-4 is easily tailored to the needs of each organization and individual using natural language. However, industry-specific versions that provide granular detail on the specific business must integrate other types of data specific to the company and ecosystem, such as operational data, scientific data, medical data, risk data, financial data, inventory, logistics, etc.

By integrating industry-specific and business-specific data into DANA (our digital assistant) plus relevant analytics and reports tailored to each individual, the KOS essentially becomes a custom EAI OS at a small fraction of the cost of stand-alone systems designed and built by each company. ?

KYield has benefited from decades of R&D. Working with hundreds of organizations, many of which are high-performance market leaders, the KOS has been refined for efficiencies. It avoids redundancy inherent in one-off system designs built by each individual organization.

?System integration plan

The integration plan identifies, defines and maps the subsystems to be integrated into the KOS so it works as a seamless, unified system. The plan includes the participants, roles and responsibilities of the integration, and lists the financial, human and technical resources necessary to execute the plan, complete with timeline and budget.

Secure access is managed in the same manner as the universal KOS—through the CKO app. When integrating sensitive legal, financial and technical data into the KOS, security becomes paramount. However, relevant data is necessary to optimize the KOS for the entire enterprise, including for specialists and managers.

One challenge is interoperability. Although APIs have become common, if the integrated apps do not support data standards and/or contain insufficient descriptive data, a conversion process is necessary. In a minority of cases integration may not be possible, the cost of integration or data conversion may be prohibitive, or negatively impact data quality.

Each subsystem integration must be individually researched, planned, executed and monitored. Depending on the organization, KYield may be able to provide full integration in collaboration with customers whereas in other cases third-party system integrators may be preferred.

Evolve or Die

No exceptions, but some have more protection than others

LLM hype notwithstanding, the AI cat is out of the bag. It has a seemingly infinite appetite. Like it or not, we are in the midst of one of the largest tech arms races in history. Even if we deeply discount the $1tn capex estimate by Goldman Sachs for AI infrastructure, investments already made will significantly influence the economy, particularly when combined with investments by other industries and governments.

Given the evidence, the question facing owners, CEOs, boards, and management teams is not whether to invest in AI, but rather what combination of buy vs. build makes the most sense for each business.

Based on three decades of R&D, KYield confirms that very few organizations have the talent and capital to build a competitive enterprise AI system. Indeed, there is significant empirical evidence of internal sabotage due to fear and related NIHS (not invented here syndrome) regarding AI systems development.

While the actual risk of being left behind is a serious concern, the fear of missing out (FOMO) may be a greater risk in the LLM arms race today, causing serious concerns.

We advise business leaders to remain grounded amidst the intense pressure and resist engaging in a self-destructive arms race. Enterprise AI decisions should be made from an evidenced-based systems perspective that is strategic and operational in nature.

“If such expectations are to be met,?ai?tools need to improve quickly, and businesses need to adopt them en masse. For the many companies along the?ai?supply chain, the stakes are getting uncomfortably high.” – The Economist.

Business leaders should consider many factors in AI adoption, including:

1. A current SWOT analysis
2. A realistic appraisal of internal AI capabilities
3. Security and safety
4. Sovereignty (lock-in and over-dependence)
5. Cyber, IP, legal and systemic risks
6. Affordability and sustainability
7. Trustworthiness of vendors and partners
8. New opportunities offered by AI systems
9. Ease of use
10. Adaptability

We acknowledge the KOS isn’t for everyone. In fact, we don’t offer our systems to about a third of the global economy due to ethics and security concerns. If the KOS became ubiquitous, like LLMs, it would no longer offer a competitive advantage. KYield was never intended to be the largest. Our goal for the KOS is to be the wisest.