Over the past several months, I have noticed an increasing number of conversations with thought leaders, professors, data scientists, solution providers, consultants, executives, and practitioners.
At first glance, they appear to have little in common.
Different industries.
Different specialties.
Different perspectives.
Yet the conversations almost always converge around the same question:
Why do organizations with access to the same technology achieve dramatically different outcomes?
That is what I find most interesting.
Not because the participants agree.
They don’t.
What is interesting is that they are increasingly arriving at similar conclusions from very different directions.
Since 1998, with funding from the Canadian Government’s Scientific Research & Experimental Development (SR&ED) program, I have worked with advanced algorithms, early AI systems, and agent-based models. That work became the foundation for Strand Commonality™ and Implementation Physics™. It also produced the Metaprise™ Model — a framework I developed in the late 1990s and early 2000s to explore the interaction of human and non-human agents, decades before the current Agentic AI discussion emerged.
Over the past year, I have overseen and participated in extensive multimodel evaluation across a wide range of AI platforms — comparing their conclusions, strengths, limitations, and failure patterns against a record I have published openly since 2007 and that consolidates documented client work, lectures, and articles reaching back to 1998 — nearly three decades of contemporaneous observation, gathered in one place rather than created there. The point was never to crown a “best model,” but to see where they converge and where they break.
One conclusion continues to emerge.
The answer to why as many as 95% of generative-AI pilots stall short of production is the same answer as to why previous technology breakthroughs consistently failed to deliver their expected outcomes: the conditions into which they were dropped did not change.
The technology is rarely the determining variable.
The determining variables are usually found elsewhere — in governance, decision rights, organizational readiness, incentive structures, leadership alignment, and the interaction between the people, processes, and systems expected to absorb the change. An important clarification: this is not traditional system or process mapping based on what the organization believes its current operating reality to be. Familiar mapping is insufficient precisely because it does not account for factors like the incentive misalignment between internal and external agents — the conditions it was never designed to see.
This is not a new observation.
It is a pattern that has repeated itself across ERP, e-Procurement, Outsourcing, Digital Transformation, Cloud Computing, and now Agentic AI.
What follows is not speculation.
It is not opinion.
It is the product of a time-stamped, living record of contemporaneous observations, documented outcomes, and verifiable case references collected across multiple technology eras.
The names of the technologies have changed.
The underlying patterns have not.
That is why these conversations continue to occur.
Not because anyone has all the answers.
But because the questions continue to matter long after the technology headlines have faded.
Truth Is Believing.
Accuracy Is Knowing.
Jon Hansen is the creator of Implementation Physics™, a research-based framework developed over nearly three decades to explain why technology initiatives succeed or fail regardless of the technology being deployed. His work spans six technology generations — from ERP through Agentic AI — and includes the Metaprise™ model first articulated in the late 1990s. His research forms the foundation for the Hansen Method™, Hansen Fit Score™ (HFS™), Phase 0™ Readiness Assessment, and the ARA™ RAM 2025™ multimodel verification architecture. He currently serves as a Board Member of the CIPS Americas Chapter.
-30-
tcummins
June 21, 2026
Yes Jon, we are moving in the same direction. It’s the management system, not the people.
piblogger
June 21, 2026
Tim,
I think that’s an important distinction.
Across multiple technology eras, I’ve found that people are often responding rationally to the conditions created by the management system around them.
Whether it was the DND project in 1998, ERP implementations, outsourcing initiatives, or today’s AI deployments, the recurring pattern is that outcomes are frequently attributed to individuals when the determining variables are actually found in decision rights, incentives, governance structures, operating assumptions, and the alignment—or misalignment—of internal and external agents, both human and agentic.
In that respect, the management system becomes the environment through which those agents interact and ultimately determine the outcome.
The technology changes.
The management-system challenge remains remarkably consistent.
Thank you for continuing to push the conversation in that direction.