By Jon W. Hansen | Procurement Insights
A recent Gartner discussion on scaling agentic AI frames the challenge largely through the lens of CIO leadership. The argument is understandable. CIOs control the infrastructure, architecture, and technology governance layers that make large-scale AI deployment possible.
But the framing felt familiar.
Not because it is wrong. Because we have heard it before — and documented what happens when it is treated as sufficient.
The Pattern the Archive Has Been Recording
During the ERP era of the late 1990s and early 2000s, the prevailing guidance was nearly identical: enterprise transformation required CIO sponsorship, strong integration discipline, and careful management of fragmented deployments. Organizations that treated ERP as an IT project rather than an enterprise transformation initiative were warned that they risked costly failures.
Most of them proceeded anyway. And most of them paid the price.
I was in those rooms. Forty years in high-technology environments — across mainframe, client-server, ERP, e-procurement, cloud, and now AI — produces a kind of pattern recognition that no single-cycle observer can replicate. What I observed then, and what the Procurement Insights archive has been documenting since 2007, is that the technology ceiling is rarely the constraint. The organizational readiness floor is.
The implementation teams were technically capable. The platforms were functionally sound. What was missing — in almost every failure I witnessed — was the cross-functional governance architecture that determined whether the technology would be absorbed productively or routed around quietly. The CIO could deploy the system. Only the organization could create the conditions for it to work.
What 2007 Already Knew
This is not retrospective wisdom. It was being documented in real time.
In August 2007 — nearly two decades ago — the Procurement Insights archive published an examination of early e-procurement initiative failures. In that post, NIGP’s Rick Grimm described the type of partnership required for successful technology adoption.
The participants he listed were not limited to IT leadership. They included the CIO, purchasing leadership, finance, major user departments, and the budgeting arm — all thinking strategically together about value and outcomes.
His conclusion when that participation was absent was straightforward:
“When you don’t have that, purchasing becomes an afterthought.”
The result, then as now, was a startlingly consistent statistic: between 75 and 85 percent of technology initiatives failed to achieve their intended outcomes.
That observation was made during the era of early e-procurement systems.
Nearly two decades later, we are discussing autonomous agents, large language models, and agentic orchestration across enterprise processes.
The structural challenge looks remarkably familiar.
And the archive did not stop documenting it in 2007.
In 2011 — four years after that e-procurement examination — a presentation titled “The Changing Face of Procurement: The End of Functional Silos” was delivered to procurement professionals. It documented what CIOs themselves were already struggling with at that time: the need to be more innovative, major challenges with a redefined role, poor existing resource utilization, and — most precisely — “looking beyond the technology to user and employee engagement.” The rationalization of existing platforms was already identified as a pressing challenge. So was the information-to-innovation transformation gap.
Every one of those CIO challenges from 2011 is present in the agentic AI conversation today. The vocabulary has changed. The underlying organizational condition has not. What that 2011 presentation was documenting — and what the archive has been recording across every technology era since — is that CIOs have never been fully equipped to resolve the human and organizational dimensions of enterprise technology adoption on their own. That is not a criticism of CIOs. It is a structural observation about where the governance gap actually lives.
Gartner’s 2026 prescription — CIO leadership as the answer to agentic AI adoption — does not engage with the dimension that both the 2007 and 2011 archive evidence identifies as the persistent failure point. The problem has been renamed. It has not been solved.
Why Agentic AI Makes the Governance Gap Larger, Not Smaller
Agentic AI does not operate inside the IT boundary. It operates across the entire organizational system. Procurement decisions intersect with finance. Supplier engagement intersects with operations. Risk management intersects with compliance and legal oversight.
These are behavioral, cultural, and process conditions that no single executive function — including the CIO — controls unilaterally.
What makes this era different from ERP and e-procurement is not the complexity of the technology. It is the speed at which the governance gap can produce damage. In 1998, a failed ERP implementation took eighteen months to surface. An agentic AI deployment making autonomous procurement decisions without the organizational readiness conditions to govern it can produce compounding failures in weeks.
The faster the technology moves, the more costly the governance shortcut becomes.
The visibility problem compounds this further. A survey of CIOs on app usage within their own organizations produced a telling result: most CIOs believed there were approximately 100 apps in use. The actual number was more than 950. That gap — between what leadership believes is deployed and what practitioners have actually built around official systems — is the governance gap made measurable. Every shortfall in the sanctioned solution became an app. If you have a gap, APP IT. The result is an organizational technology reality that no single executive, including the CIO, fully sees — and into which agentic AI is now being deployed.
This does not diminish the importance of CIO leadership. CIOs play a critical role in ensuring that the technical architecture supporting AI systems is secure, scalable, and well governed. But the historical pattern — documented across forty years and three major technology eras in this archive — is unambiguous.
CIO leadership is necessary.
It has never been sufficient.
What Sufficient Actually Looks Like
Every major enterprise technology era has eventually arrived at the same realization: successful adoption requires cross-functional governance structures that align decision rights, incentives, and accountability across the organization.
When that alignment is absent, technology capabilities expand while outcomes stagnate. The implementation looks successful at go-live. The results tell a different story eighteen months later.
The question Gartner’s CIO-centric framework does not ask — and that agentic AI cannot answer for itself — is whether the human systems it is being deployed into are prepared to absorb it productively or will route around it quietly.
That is not a technology question. It is a governance question that belongs at the full C-suite and board level. Phase 0™ exists to answer it — before the deployment decision, while the outcome can still be shaped.
The technology stack evolves quickly. The organizational conditions required to absorb it tend to change much more slowly.
And that, perhaps more than anything else, explains why each new generation of enterprise technology arrives with extraordinary promise — and why so many organizations continue to struggle converting that promise into sustained results.
The Value of Institutional Memory
One of the advantages of maintaining a longitudinal archive is that it allows patterns like this to remain visible across technology eras. What appears new in one cycle often echoes lessons documented in another.
Institutional memory does not eliminate failure. But it does help explain why the same questions keep returning — and why the organizations that ask them before deployment consistently outperform those that ask them after.
The 2007 article referenced above is available here: https://procureinsights.com/2007/08/01/technology-and-the-growing-talent-crunch/
The Procurement Insights archive — 3,300+ published documents spanning eighteen years — is available at procureinsights.com. The Hansen Fit Score™, Hansen Method™, Phase 0™, and RAM 2025™ multimodel validation framework are proprietary frameworks of Hansen Models™.
Hansen Website and Resources: https://hansenprocurement.com/
Jon W. Hansen is the founder of Hansen Models™ and has been publishing independently since 2007 — no vendor sponsorships, no referral arrangements.
-30-
Procurement Insights
CIO Leadership Is Necessary. It Has Never Been Sufficient.
Posted on March 16, 2026
0
By Jon W. Hansen | Procurement Insights
A recent Gartner discussion on scaling agentic AI frames the challenge largely through the lens of CIO leadership. The argument is understandable. CIOs control the infrastructure, architecture, and technology governance layers that make large-scale AI deployment possible.
But the framing felt familiar.
Not because it is wrong. Because we have heard it before — and documented what happens when it is treated as sufficient.
The Pattern the Archive Has Been Recording
During the ERP era of the late 1990s and early 2000s, the prevailing guidance was nearly identical: enterprise transformation required CIO sponsorship, strong integration discipline, and careful management of fragmented deployments. Organizations that treated ERP as an IT project rather than an enterprise transformation initiative were warned that they risked costly failures.
Most of them proceeded anyway. And most of them paid the price.
I was in those rooms. Forty years in high-technology environments — across mainframe, client-server, ERP, e-procurement, cloud, and now AI — produces a kind of pattern recognition that no single-cycle observer can replicate. What I observed then, and what the Procurement Insights archive has been documenting since 2007, is that the technology ceiling is rarely the constraint. The organizational readiness floor is.
The implementation teams were technically capable. The platforms were functionally sound. What was missing — in almost every failure I witnessed — was the cross-functional governance architecture that determined whether the technology would be absorbed productively or routed around quietly. The CIO could deploy the system. Only the organization could create the conditions for it to work.
What 2007 Already Knew
This is not retrospective wisdom. It was being documented in real time.
In August 2007 — nearly two decades ago — the Procurement Insights archive published an examination of early e-procurement initiative failures. In that post, NIGP’s Rick Grimm described the type of partnership required for successful technology adoption.
The participants he listed were not limited to IT leadership. They included the CIO, purchasing leadership, finance, major user departments, and the budgeting arm — all thinking strategically together about value and outcomes.
His conclusion when that participation was absent was straightforward:
“When you don’t have that, purchasing becomes an afterthought.”
The result, then as now, was a startlingly consistent statistic: between 75 and 85 percent of technology initiatives failed to achieve their intended outcomes.
That observation was made during the era of early e-procurement systems.
Nearly two decades later, we are discussing autonomous agents, large language models, and agentic orchestration across enterprise processes.
The structural challenge looks remarkably familiar.
And the archive did not stop documenting it in 2007.
In 2011 — four years after that e-procurement examination — a presentation titled “The Changing Face of Procurement: The End of Functional Silos” was delivered to procurement professionals. It documented what CIOs themselves were already struggling with at that time: the need to be more innovative, major challenges with a redefined role, poor existing resource utilization, and — most precisely — “looking beyond the technology to user and employee engagement.” The rationalization of existing platforms was already identified as a pressing challenge. So was the information-to-innovation transformation gap.
Every one of those CIO challenges from 2011 is present in the agentic AI conversation today. The vocabulary has changed. The underlying organizational condition has not. What that 2011 presentation was documenting — and what the archive has been recording across every technology era since — is that CIOs have never been fully equipped to resolve the human and organizational dimensions of enterprise technology adoption on their own. That is not a criticism of CIOs. It is a structural observation about where the governance gap actually lives.
Gartner’s 2026 prescription — CIO leadership as the answer to agentic AI adoption — does not engage with the dimension that both the 2007 and 2011 archive evidence identifies as the persistent failure point. The problem has been renamed. It has not been solved.
Why Agentic AI Makes the Governance Gap Larger, Not Smaller
Agentic AI does not operate inside the IT boundary. It operates across the entire organizational system. Procurement decisions intersect with finance. Supplier engagement intersects with operations. Risk management intersects with compliance and legal oversight.
These are behavioral, cultural, and process conditions that no single executive function — including the CIO — controls unilaterally.
What makes this era different from ERP and e-procurement is not the complexity of the technology. It is the speed at which the governance gap can produce damage. In 1998, a failed ERP implementation took eighteen months to surface. An agentic AI deployment making autonomous procurement decisions without the organizational readiness conditions to govern it can produce compounding failures in weeks.
The faster the technology moves, the more costly the governance shortcut becomes.
The visibility problem compounds this further. A survey of CIOs on app usage within their own organizations produced a telling result: most CIOs believed there were approximately 100 apps in use. The actual number was more than 950. That gap — between what leadership believes is deployed and what practitioners have actually built around official systems — is the governance gap made measurable. Every shortfall in the sanctioned solution became an app. If you have a gap, APP IT. The result is an organizational technology reality that no single executive, including the CIO, fully sees — and into which agentic AI is now being deployed.
This does not diminish the importance of CIO leadership. CIOs play a critical role in ensuring that the technical architecture supporting AI systems is secure, scalable, and well governed. But the historical pattern — documented across forty years and three major technology eras in this archive — is unambiguous.
CIO leadership is necessary.
It has never been sufficient.
What Sufficient Actually Looks Like
Every major enterprise technology era has eventually arrived at the same realization: successful adoption requires cross-functional governance structures that align decision rights, incentives, and accountability across the organization.
When that alignment is absent, technology capabilities expand while outcomes stagnate. The implementation looks successful at go-live. The results tell a different story eighteen months later.
The question Gartner’s CIO-centric framework does not ask — and that agentic AI cannot answer for itself — is whether the human systems it is being deployed into are prepared to absorb it productively or will route around it quietly.
That is not a technology question. It is a governance question that belongs at the full C-suite and board level. Phase 0™ exists to answer it — before the deployment decision, while the outcome can still be shaped.
The technology stack evolves quickly. The organizational conditions required to absorb it tend to change much more slowly.
And that, perhaps more than anything else, explains why each new generation of enterprise technology arrives with extraordinary promise — and why so many organizations continue to struggle converting that promise into sustained results.
The Value of Institutional Memory
One of the advantages of maintaining a longitudinal archive is that it allows patterns like this to remain visible across technology eras. What appears new in one cycle often echoes lessons documented in another.
Institutional memory does not eliminate failure. But it does help explain why the same questions keep returning — and why the organizations that ask them before deployment consistently outperform those that ask them after.
The 2007 article referenced above is available here: https://procureinsights.com/2007/08/01/technology-and-the-growing-talent-crunch/
The Procurement Insights archive — 3,300+ published documents spanning eighteen years — is available at procureinsights.com. The Hansen Fit Score™, Hansen Method™, Phase 0™, and RAM 2025™ multimodel validation framework are proprietary frameworks of Hansen Models™.
Hansen Website and Resources: https://hansenprocurement.com/
Jon W. Hansen is the founder of Hansen Models™ and has been publishing independently since 2007 — no vendor sponsorships, no referral arrangements.
-30-
Share this:
Related