Your blog post

FIELD NOTES | JUNE 2026

AI Can't See Your Organization. You Can.

Why Organization Development must lead AI implementation — and what happens when it doesn't.

By Faith Addicott, Executive Director, Good Work Collective, SPC

In 2025, global enterprises invested $684 billion in AI initiatives. By year-end, more than $547 billion of that investment had failed to deliver intended business value. Executives are mandating AI adoption. Vendors are selling it. Technology teams are deploying it. HR leaders are racing to restructure around it. And the organizations on the receiving end of all that velocity are, in most cases, not understood well enough to receive it.

The dominant frameworks for AI adoption see the org chart. They do not see the organization. They map roles, functions, and workflows, then design agent architectures around them. What they miss is the living system underneath: the flows of information, power, relationship, and meaning that actually hold organizations together. Those flows do not appear on any org chart. They live in the lines between the boxes, and that is precisely where AI is being deployed blind.

Organizations preparing for AI implementation should be hiring Organization Development practitioners as human systems architects — before, during, and after any AI deployment touches the organization. Before engagement collapses. Before trust is broken. Before the knowledge walks out the door.

Obviously this becomes tricky when the speed of adoption is presented as a mandate for jumping off a cliff without worrying about consequences, which seems be the tone of much AI chatter at the moment. AI tech bros are presenting rapid, unstudied implementation as a done deal, pushing questions of ethics, veracity and ROI under the shiny new rug of progress. If we can’t slow down (much) it becomes even more vital that implementation is paired with actual organizational knowledge.

AI can do tasks. AI can fill roles. AI cannot supplant the information, knowledge, and power that occupies the lines between the boxes on an org chart.

The Crisis Nobody Is Naming

Gallup's 2026 State of the Global Workplace report puts global employee engagement at 20% — the lowest level since pandemic lockdowns, and the second consecutive year of decline. The economic cost of that disengagement exceeds $10 trillion annually, equivalent to 9% of global GDP. Ten trillion dollars, every year, because people are showing up to work and not bringing themselves with them.

The most striking finding in that report is not the aggregate number. Workers in organizations actively implementing AI are more likely to believe their jobs will be eliminated within five years than the general workforce. AI implementation, as currently practiced, is accelerating the very disengagement it is supposed to solve.

The financial picture confirms it. 88% of AI pilots never reach production at all. Of those that do, MIT research found that 95% of generative AI pilots show zero measurable bottom-line impact. Large enterprises abandoned an average of 2.3 AI initiatives in 2025, at a sunk cost of $7.2 million each. Completed-but-failed projects returned a -72% ROI on average.

Across more than 2,400 enterprise AI initiatives, researchers consistently find the same root cause: poor alignment, absent change infrastructure, and technology deployed into human systems that were never assessed or prepared for what was being asked of them. The organizations beating those odds invest 30 to 40% of their transformation budgets in change management. Most organizations invest 10%. The gap is not a budget gap. It is a discipline gap — and OD is the discipline.

What the Frameworks Get Right — and What They Miss

Josh Bersin's HR 2030 vision, launched this month, is the most comprehensive framework available for understanding what agentic AI means for HR. The analysis is rigorous, the architecture sophisticated, and the case it makes for integrated, data-driven people systems is sound. At its foundations, though, it is an optimization of the system as it already exists. It takes the org chart as given, designs agents for the boxes, and does not ask what lives in the lines between them — or what happens to those lines when you route work, decisions, and relationships through an AI layer the organization was never designed around.

Bill Jennings' forthcoming book Humanity: Leading the Powershift makes the essential corrective argument: frameworks like Bersin's are so corporate-centered that they produce harm even while improving efficiency metrics. Jennings calls for what he names Taijitu Design Principles — the commitment to enhance each individual's ability to create and control their own destiny must be equal to and balanced with ensuring the company can do the same. He is right about the gap. Both frameworks, though, stop one discipline short of naming who can actually close it.

Identifying what HR should become is critical work. Identifying the rights workers need in the AI era is critical work. Building the organizational conditions in which any of that can actually happen requires OD.

Bersin architects the agentic HR future. Jennings argues it must be workforce-centered. Both are right. What neither framework supplies is the discipline trained to read the organizational system as a living whole.

The Org Chart Is a Fiction. The Organization Is Real.

Every organization has two structures. The first is the formal one — the org chart, the reporting lines, the defined roles and accountabilities. The second is the actual organization: the informal networks of trust and distrust, the shadow hierarchies of influence, the gatekeepers who control information not because they are empowered to but because they always have been, the teams that collaborate effectively across department lines, the teams that are theoretically aligned and functionally at war. The second structure never appears on any org chart, never surfaces in any HRIS, and never shows up in any agent architecture. It is, however, the organization that will actually implement your AI strategy.

When Bob in accounts payable leaves, the organization does not lose someone who processed invoices. It loses the person who knew which vendors to call first, who had the informal relationship with the CFO's assistant that moved emergency approvals through, who caught the discrepancies that saved the company $40,000 last year because he had been watching the same numbers long enough to recognize when something was wrong. Invoice processing can be automated. What Bob actually was in that organization cannot.

The data on what happens when organizations miss this is stark. 42% of institutional knowledge resides solely with individual employees and is never shared with coworkers. IDC estimates Fortune 500 companies lose $31.5 billion annually from knowledge-sharing failures alone. Two-thirds of companies that conducted AI-driven layoffs in the past year are already rehiring the roles they eliminated. Only 8.4% report that the restructuring delivered on its promises. Forrester projects that half of all AI-attributed layoffs will be quietly reversed by 2027.

When you eliminate the account manager who carries ten years of relationship context, or the QA engineer whose institutional knowledge is the only thing catching defects before they reach customers, you are not cutting fat. You are cutting load-bearing walls.

— Unite.ai, 2026

What OD Practitioners Actually Do

Organization Development is the applied science of organizational systems — the study and practice of understanding, diagnosing, and intervening in the patterns of behavior, power, communication, culture, and structure that make organizations what they are. OD practitioners are trained to see not just what an organization says it is, but what it actually is: how decisions really get made, where information really flows, who holds real influence, and what the culture actually rewards and punishes, regardless of what the values statement says.

HR manages people within the system. Change management deploys specific interventions within the system. OD architects the system itself. In an AI implementation context, that means asking questions that no technology vendor and no HR framework is currently asking:

• What are the power dynamics this change will disrupt, and who will fight it, and why?

• What informal communication channels does this process depend on that the AI system will not replicate?

• What psychological safety conditions need to be in place for people to use this tool honestly rather than performatively?

• What does success look like in 18 months, and how will we know if the human system has adapted well or is quietly failing?

These are the questions that determine whether AI implementation produces the value it promises — or produces a faster, more expensive version of the dysfunction that was already there.

The Organization as Relationship in Pursuit of Good Work

An organization is not, at its core, a structure. It is a series of agreements and relationships through which people pursue together what they cannot achieve alone. That is its fundamental nature, and it is why organizational life is irreducibly human, even as AI transforms what humans do within it.

AI can do tasks, often faster, more accurately, and at greater scale than any human. That is genuinely valuable. And it is genuinely limited. AI can fill roles, in the sense that it can perform the defined outputs associated with a role. What it cannot do is occupy the relational position a human being holds in a living organizational system. It cannot hold the trust someone earned over a decade. It cannot read the room when a team is demoralized. It cannot be the person someone calls when they do not know what to do. Those capacities are not soft features of organizational life. They are its connective tissue.

Organizations are constituted by their flows of information, power, and relationship — not by the roles that formally channel those flows. The boxes on an org chart are interchangeable. The relationships, trust networks, and informal power structures that actually move work through an organization are not. They have to be understood, respected, and deliberately built for, especially when change is coming fast.

What We Are Building — and the Invitation

Good Work Collective is partnering with AI Monster and their MonsterSphere platform to conduct the first field-defining research study on Human-AI Change in organizations from an OD perspective. Launching at OD Out Loud Revival 2026 — October 25-28, Kansas City, Kansas — the study will produce an open-access research paper, a practitioner playbook including an OD-grounded AI readiness assessment, and a living knowledge base on MonsterSphere that evolves as practitioners contribute real case experience over time.

MonsterSphere is built on the premise that the wisdom needed to navigate AI responsibly is not in the models. It is in the people who have already encountered AI in real organizational contexts, learned what works and what fails, and can share that knowledge in ways that are specific, honest, and attributable. The partnership with GWC exists because that platform needed OD not just as a content area, but as the discipline that sets the terms for what good implementation knowledge actually looks like. Having OD at the table where those standards are being built, rather than brought in after the architecture is already decided, is the point.

If you are an OD practitioner already thinking about this work and want to be part of the research, the Revival, or the practitioner community forming around it, we want to hear from you. And if you want to go deeper on this argument before October, join us for a free 90-minute virtual session in August. Registration opens soon.

AI Can't See Your Organization. You Can.

A free 90-minute virtual session for OD practitioners | August 2026

Hosted by Good Work Collective x AI Monster / MonsterSphere

Registration opens soon: goodworkcollective.net

Faith Addicott is the Executive Director of Good Work Collective, SPC, a cooperatively owned Organization Development consulting practice, and the producer of OD Out Loud Revival 2026. She writes and consults at the intersection of organizational systems, cooperative ownership, and the future of work.

Connect

Join us to uplift OD practice together

Email

info@goodworkcollective.net

© 2025. All rights reserved.