The Silent Partner.

Leaders experienced the tool as a silent partner, not a script generator.

Across post-intervention interviews, leaders primarily described the tool as a preparatory aid that helped them think through conversations in advance rather than as a source of verbatim language. The effect was metacognitive before it was behavioral — a shift in how leaders thought about an approach before it became a change in how they acted.

Awareness before action is consistent with Day & Harrison's (2007) developmental sequence. — Discussion, p. 7

Individualized Consideration emerged as the central dimension.

Of the four transformational dimensions, Individualized Consideration was most visibly scaffolded. Leaders used the tool to tailor communication, anticipate reactions, and prepare for more developmental conversations. P-10 built a structured scorecard and roadmap, replacing what they called motherhood statements with specific behavioral expectations.

Several cases suggested activation of less-developed TL dimensions.

Both P-05 and P-06 identified Intellectual Stimulation as their least-developed dimension pre-intervention and later described post-intervention movement. A team member's unsolicited comment to P-06 — "Where are these questions coming from now?" — offered the strongest external corroboration in the dataset.

Engagement patterns were linked to distinct outcome profiles.

Frequency alone did not explain difference in reported outcomes. Integration rhythm mattered. A five-pattern typology (see Table II below) distinguished frequent-light integration from deadline-reactive use — and produced qualitatively different kinds of reported change.

Framework separation between administrative and developmental intent.

Several participants described using the tool to disentangle the institutional review process from the developmental conversation within it. P-08 described the tool as a "sounding board" that facilitated this separation — a potentially valuable function in regulated environments where administrative compliance and developmental depth compete.

Domain-specific category-blindness limited uptake — even among AI-fluent leaders.

A distinct pattern in the dataset was that several AI-fluent leaders had not previously considered AI relevant to leadership work until the study explicitly framed it that way. P-07, whose role was the most AI-focused in the sample, reported the study surfaced a use case that had not previously occurred to them despite deep professional immersion in the technology. P-10, a significant prior AI user, described the non-use for leadership as a failure of my imagination rather than a deficiency in the tools.

Adoption barriers, then, may stem not only from trust or access — but from whether leaders can imagine AI as relevant to leadership work in the first place. General AI fluency predicts engagement only once the use case has been admitted to the leader's mental category of AI-relevant work.

A failure of my imagination — not a deficiency in the tools. — P-10, post-intervention

Use extended beyond one-to-one downward coaching.

Participants extended the tool into upward influence (P-03 with their own manager), lateral influence (P-06), team-development artifacts (P-04's distributed analyst-uplift playbook), second-level coaching cascades (P-11), and audience simulation for testing alternative phrasings.

Not all reported shifts were tool-driven.

Some changes appeared connected to broader role demands, organizational pressure, or accumulated leadership experience during the study period. P-07's substantial shift was linked primarily to the realities of scaling a rapidly growing function, not the tool. The tool should be understood as one possible contributor to leadership development — not its sole source.

Governance is the condition, not the caveat.

Participants raised data governance questions before discussing the tool's value. Who can see my prompts? How is data stored? Could this appear in personnel files? Without technologically guaranteed psychological safety from institutional surveillance, engagement remained superficial. In regulated settings, autonomy-supportive design cannot function if the governance substrate does not.