Most organizations concentrate their change management efforts around deployment and post-implementation support. That’s a strategic miscalculation. The period before AI is introduced is where resistance is seeded—and where organizational support must be cultivated. By the same time a system is live, employee perceptions are already set.
Organizational Support Theory (OST), introduced in Part 1, explains this dynamic precisely. Employees develop their perception of organizational support (POS) well before any formal rollout. They observe how leadership communicates. They notice whether their concerns are solicited or dismissed. They read the cultural signals. If those signals communicate “this is happening to you, not with you,” resistance follows —predictably and at scale [3].
Organizations that wait until go-live to manage the human side of change have already lost the window that matters most.
Prosci’s ADKAR model —Awareness, Desire, Knowledge, Ability, and Reinforcement —is one of the most widely adopted individual change frameworks in organizational practice [4]. Originally developed as a sequential model for how individuals move through change, ADKAR provides a precise roadmap for building organizational support during the critical pre-launch phase.
Unlike frameworks that treat change as a top-down directive, ADKAR is grounded in the lived experience of individual employees navigating uncertainty. That focus is exactly what’s missing from most AI rollouts.
Awareness: Do Employees Know What’s Coming—and Why?
The first building block isn’t simply announcing that AI is on the way. It’s building genuine understanding of the need for change. Research shows that when organizations communicate a clear AI implementation plan, employees are 4.7 times more likely to feel comfortable using AI in their roles. Yet only 15% of organizations provide that clarity [1].
Awareness, as ADKAR defines it, isn’t a memo or a town hall. It’s a sustained, multi-channel communication effort that connects the AI initiative to organizational goals, competitive realities, and —critically —the implications for individual roles. Employees aren’t just asking “what is this?” They’re asking, “what does this mean for me?” Failing to answer that question directly leaves a vacuum that anxiety and rumor readily fill.
Desire: From Compliance to Commitment
Awareness without desire produces compliance, not commitment —and compliance is fragile. The moment the pressure of a launch fades, passive resistance moves in to take its place.
Desire is built through involvement. When employees have a meaningful role in shaping how AI tools are deployed —through pilot programs, cross-functional working groups, or structured feedback channels —their relationship to the change shifts. They become participants rather than bystanders. OST research consistently finds that employees who perceive the organization as valuing their input reciprocate with greater commitment to organizational goals [3]. Desire is where perceived organizational support is most directly cultivated —or destroyed.
Prosci’s longitudinal research identifies employee resistance and lack of visible executive sponsorship as the two most frequently cited barriers to successful change [5]. Both are desire problems —and both are preventable if addressed in the pre-implementation window.
Knowledge: Functional Literacy Before Day One
Knowledge, in the ADKAR framework, refers to understanding how to change —not simply knowing that change is coming. In AI implementation, pre-launch knowledge-building means equipping employees with enough conceptual grounding to engage with the technology meaningfully, not just to comply with a mandate.
The distinction matters. A workforce that understands what an AI system can and cannot do —and where human judgement remains irreplaceable —is far more likely to integrate those tools effectively. Conversely, a workforce that arrives at launch day without that foundation will default to avoidance or misuse. As Adaptive Structuration Theory established in Part 1, people will adapt technology to their own needs whether or not those adaptations align with organizational intent.
Pre-implementation knowledge-building doesn’t require deep technical training. It requires functional literacy—a working understanding of the AI’s purpose, its constraints, and its relationship to existing workflows.
Ability: Identifying the Gap Before It Becomes a Problem
Ability— the capacity to actually implement the change —is typically treated as a post-launch concern. But in most AI rollouts, the disparity between knowledge and ability is predictable in advance. Organizations that identify high-risk populations early - and those whose roles will be most disrupted, those with the lowest digital fluency - and begin targeted support planning before launch are far better positioned to undertake complications before they arise.
This is where OST does its most direct work in pre-implementation. Employees who believe the organization is investing in their success —not just the success of the technology — report significantly higher levels of readiness and engagement [3]. Ability is less about technical proficiency than it is about confidence, and confidence is built through demonstrably supportive organizational behavior, before the pressure of go-live arrives.
Reinforcement: Building the Infrastructure Before Launch
Reinforcement is the ADKAR element most associated with sustaining change after implementation. But the groundwork for reinforcement must be established before the system goes live. Organizations that define success metrics, feedback loops, and recognition mechanisms during the pre-launch phase build the infrastructure needed to sustain adoption once the transition occurs.
Without that groundwork, post-launch reinforcement becomes reactive — a scramble to recover adoption rates that were never fully established.
No pre-implementation strategy succeeds without visible, credible leadership sponsorship. Prosci’s research consistently identifies lack of active and visible leadership as the primary driver of change failure [5]. In AI implementation, this challenge is compounded by a reality that few organizations acknowledge openly: leaders themselves are often uncertain about AI’s implications. That uncertainty, left unaddressed, becomes a cultural signal.
Leaders who model intellectual curiosity, communicate transparently about both the opportunity and the disruption AI represents, and acknowledge what remains unknown do more to build organizational support than any formal program. In this context, authenticity isn’t a soft skill—it’s a strategic asset.
Applied to the pre-implementation window, ADKAR reframes what building organizational support actually means. It isn’t a communication campaign. It isn’t a training schedule. It’s a structured, sequential process of moving employees from unawareness to readiness—before the technology is ever introduced.
The organizations that get AI implementation right aren’t those with the most sophisticated systems. They’re the ones that treated the human side of implementation with the same rigor applied to the technical side. As established in Part 1, AI implementation is a sociotechnical transformation. ADKAR provides a roadmap for the social half of that equation—and it begins well before the switch is flipped.
Part 3 of this series will explore specific OD interventions for each phase of AI implementation—how to design AI-augmented work that serves both organizational and human needs simultaneously.
The following steps operationalize the ADKAR principles outlined above. These are intended for future application and planning purposes.
References
[1] Gallup. (2024). AI in the workplace: Answering 3 big questions. https://www.gallup.com/workplace/651203/workplace-answering-big-questions.aspx
[2] McKinsey & Company. (2015). Changing change management. McKinsey Quarterly. https:// www.mckinsey.com/featured-insights/leadership/changing-change-management
[3] Eisenberger, R., & Stinglhamber, F. (2011). Perceived organizational support: Fostering enthusiastic and productive employees. American Psychological Association. https://doi.org/10.1037/12318-000
[4] Hiatt, J. M. (2006). ADKAR: A model for chang in business, government, and our community. Prosci Learning Center Publications.
[5] Prosci. (2023). Best practices in change management (12th ed.). Prosci Inc.
