Organizations today are racing to embed artificial intelligence into everyday operations, yet many stumble at the very first step: deciding who will lead the charge. The concept of an AI championโ€”a person granted protected time to explore, build, and advocate for AI solutionsโ€”has emerged as a promising model. However, when leadership spreads the expectation that everyone should devote a sliver of their schedule to AI, the initiative often dissipates into vague good intentions. Without explicit carveโ€‘outs and clear accountability, the work gets squeezed by urgent tasks, and the momentum stalls. This paradox lies at the heart of many transformation efforts: genuine progress requires a dedicated owner who can focus, experiment, and demonstrate value, rather than a diffuse sense of responsibility that leaves no one truly accountable.

At the Zapier Outpost Windsor retreat in September, an operations leader shared a cautionary tale that illustrated this dynamic vividly. Their organization had agreed to shield 20โ€ฏ% of an AI championโ€™s calendar for experimentation, only to have a manager reclaim that time a month later because pressing work demanded attention. The retreat facilitator, Philip Lakin, Zapierโ€™s director of AI transformation, did not contest the decision; instead, he pointed out that the failure stemmed from a lack of genuine leadership alignment. When compensation, incentives, and calendar protections do not reflect the stated priority, the initiative remains a paper exercise. The anecdote underscores a fundamental truth: unless leaders visibly back the commitment with tangible resources, the championโ€™s role will be undermined by competing priorities.

Zapierโ€™s own internal structure offers a contrasting example of how to institutionalize the champion model effectively. Members of the AI transformation guild remain embedded in their respective business lines, yet 30โ€ฏ% to 40โ€ฏ% of their workload is officially earmarked for transformation activities. To make this possible, Zapier proportionally reduces their other duties, ensuring the protected time is real and not merely aspirational. The role itself is broken down into three concrete functions: listening to endโ€‘users for pain points, building prototypes or workflows, and translating the broader AI mandate into measurable outcomes that the team can track. This clear delineation of responsibilities helps champions stay focused and provides a framework for evaluating impact.

Conversely, attempting to distribute AI responsibility thinly across an entire workforce often backfires. Telling every employee that AI constitutes, say, 10โ€ฏ% of their job creates a diffusion of accountability where no individual feels ownership over the initiative. Without a dedicated person to drive experiments, gather data, and iterate, the effort becomes a series of isolated, lowโ€‘impact attempts that never coalesce into meaningful change. Leaders may think they are fostering a culture of experimentation, but in practice they are diluting focus and making it difficult to attribute results to any specific action. The lesson is clear: sustainable AI adoption requires concentrated effort rather than a blanket, lowโ€‘intensity expectation.

The traditional sales pitch for AIโ€”promising efficiency gains, time savings, and cost reductionsโ€”often misses the mark with the very people who would implement the technology. Frontline workers frequently interpret such messages as a threat: they fear that mapping their workflows is the first step toward automation that could replace them, or they simply lack the bandwidth to experiment amid existing workloads. This mismatch breeds skepticism and can even trigger guilt when employees do try AI tools, worrying that they are inadvertently facilitating their own obsolescence. To overcome this resistance, advocates must reframe the conversation around relief and personal benefit rather than abstract productivity metrics.

Philip Lakin advised retreat participants to highlight how AI can remove the most tedious, frustrating parts of a job or make an individual look good in front of peers and supervisors. By asking workers what they dislike about their daily tasksโ€”whether itโ€™s repetitive data entry, endless ticket triage, or cumbersome reportingโ€”champions can identify highโ€‘value automation targets that resonate on a personal level. When a solution visibly eliminates a pain point, adoption becomes selfโ€‘reinforcing: employees experience immediate relief, see tangible improvements in their work lives, and become enthusiastic advocates. This shift from โ€œsave time for the companyโ€ to โ€œmake your day betterโ€ aligns incentives and reduces fear.

Broader industry trends reinforce the workersโ€™ apprehensions. Metaโ€™s chief technology officer has warned that AIโ€‘driven productivity gains often translate into more work rather than additional leisure time, as organizations reinvest freed capacity into new initiatives. In South Korea, Hyundai workers voted to strike over fears that automation would jeopardize jobs, illustrating how perceived threats can trigger organized resistance. Surveys of younger professionals reveal a growing sense of guilt when using AI at work, stemming from concerns that they are enabling displacement of colleagues. These signals suggest that any AI rollout must address not only technical feasibility but also the human and cultural dimensions that shape employee sentiment.

Identifying the right individuals to champion AI does not require a lengthy search; it can be accelerated through targeted, lowโ€‘commitment experiments. Grady Sokaras, Zapierโ€™s head of sales, described two practical pathways: hiring a dedicated specialist outright, or running a brief workshop to gauge interest and aptitude. Zapierโ€™s internal practice involves a 90โ€‘minute session where a small cohort builds its first simple workflow. Observers note who grasps the concepts quickly, shows enthusiasm, and begins to think about automation possibilities. Those individuals are then earmarked for deeper training, with their roles reshaped so that 30โ€ฏ% to 40โ€ฏ% (or even 100โ€ฏ%) of their time is devoted to AIโ€‘focused work, backed by clear objectives and measurement.

The underlying theory behind this approach is that people who have handsโ€‘on experience building automationโ€”even if they never move those prototypes into productionโ€”develop a sharper eye for problematic processes. By engaging in the act of creation, they learn to spot inefficiencies, data bottlenecks, and opportunities for integration that might otherwise remain invisible. However, building ability alone is insufficient; innovations must be marketed internally to gain traction. As Lakin noted, โ€œThings do not, especially in operations, things do not internally market themselves.โ€ Champions therefore need to translate technical successes into stories that resonate with stakeholders, highlighting metrics, time saved, or error reductions that justify continued investment.

To safeguard a championโ€™s protected time, leaders should arm them with concrete data that demonstrates impact. Lakin urged participants to return to their managers with beforeโ€‘andโ€‘after figures: for example, showing how a manual process that once took ten hours per week now requires only two, freeing up capacity for higherโ€‘value work. When a manager can see a direct link between the championโ€™s efforts and measurable outcomes, the rationale for preserving that time becomes selfโ€‘evident. This evidenceโ€‘based dialogue transforms the conversation from a request for favor into a discussion about resource allocation driven by results.

Lakin also shared a threeโ€‘layer diagnostic framework to pinpoint why AI initiatives stall. At the base level, if frontline employees are not surfacing ideas to the AI guild, the problem lies in enablement and skillsโ€”workers may lack the confidence or knowledge to propose improvements. In the middle layer, abundant ideas that never progress due to data silos or crossโ€‘departmental friction indicate a breakdown in coordination; someone must integrate disparate ticketโ€‘triaging patterns into a unified workflow. At the top level, when both lower layers function yet nothing changes, the issue is either a vague mandate or a failure of senior leaders (VPs, directors) to translate strategic goals into measurable objectives that affect their own performance metrics and reward structures.

One retreat attendee lamented that their organization felt โ€œa million years awayโ€ from creating a role that bridges operations and IT, reflecting a common perception of entrenched silos. Kevin Forestell of Zapier countered by urging leaders to compress their timelines: given the rapid pace of AI advancement, halving the estimated time needed for such structural changesโ€”or at least holding the goal looselyโ€”can accelerate progress. He noted that the very same attendee had acknowledged that their own role had not existed nine months prior, illustrating how quickly new functions can emerge when urgency and clarity are present.

Market activity underscores the growing recognition that focused talent development is essential to bridge the AI skills gap. The AI skills firm Multiverse secured a $70โ€ฏmillion funding round in May, signalling investor confidence in structured training programs designed to produce readyโ€‘toโ€‘work AI practitioners. This influx of capital highlights a broader trend: organizations are increasingly willing to invest in upskilling, but they must couple those investments with protected time, clear metrics, and a culture that trusts and rewards demonstrated results. For leaders seeking to navigate the AI champion dilemma, the actionable path is clear: identify eager builders, grant them genuine time shields, measure and communicate impact, align incentives, and iteratively refine the organizational structure to sustain momentum.