Leadership Backs AI, But Departments Struggle to Execute

By Hou Yanjun

Despite leadership endorsement, approved budgets, and multiple training sessions, AI initiatives frequently stall at the departmental level. While everyone nods in agreement during meetings, urgent operational tasks consistently take precedence post-meeting.

This pattern often leads to assumptions about middle management's resistance or poor execution. The typical response? More meetings, deadlines, and progress tracking—demanding each department propose AI use cases.

However, there's a significant gap between a CEO's directive to "implement AI this year" and a department's understanding of which specific tasks to adjust next week. While the strategic direction is set, critical operational questions remain unanswered: Who executes? How to reprioritize existing work? Which processes can change? Who resolves issues?

When these questions linger, perceived slow progress often reflects department heads' legitimate concerns about maintaining business continuity.

New Workloads Without Reduced Expectations

Consider a sales manager's reality: Monthly targets for contracts and payments remain unchanged, while they're suddenly tasked with piloting an AI-powered client tracking tool.

Implementation requires client data reorganization, modified documentation processes, and training time—all while maintaining legacy systems. Performance metrics stay unchanged, and clients won't accommodate slower responses due to AI trials.

Faced with this reality, sales managers naturally prioritize revenue-generating activities over experimental AI benefits. Repeatedly emphasizing AI's importance proves ineffective without concrete guidance on:

• Priority clients/tasks for testing
• Available bandwidth allocation
• Which existing tasks can deprioritize
• Clear pilot adjustment/termination criteria

Resource allocation also requires nuance. Piloting a ready-made tool differs fundamentally from redesigning entire workflows—some initiatives need dedicated personnel, not just "spare time" contributions.

An October 14, 2026 ASCM-DSCI supply chain AI adoption survey (107 North American respondents) identified skills gaps and unclear ROI as top barriers—highlighting the distinction between supporting AI and having execution capabilities.

Execution Barriers Beyond Departmental Control

Some roadblocks require cross-functional solutions. For example:

• Sales teams using AI for proposals need product data from other departments
• Pricing policies require finance approval
• Delivery timelines need supply chain confirmation

Sales managers can mandate tool usage but lack authority to modify other departments' data practices or approval workflows. If manual verifications persist, even elegant tools create minimal efficiency gains.

In such cases, monitoring sales team adoption rates misses the point. Leadership must establish:

• Data ownership and update protocols
• System integration standards
• Cross-departmental decision-making authority

These represent fundamental responsibility redistributions—not solvable through project groups alone. Cross-functional initiatives require empowered coordinators and escalation pathways to prevent endless "in discussion" statuses.

Executive support must manifest in resolving inter-departmental deadlocks, not just reviewing progress reports.

Addressing Managerial Concerns Transparently

Resistance isn't always resource-related. Legitimate concerns include:

• Reduced informational advantage through AI transparency
• Team downsizing without workload reduction
• Technical knowledge gaps affecting leadership credibility

An April 2026 UK IFS working paper (2,000 US/UK managers) found exposure to "AI job displacement" narratives reduced adoption willingness—highlighting how perceptions shape attitudes.

Organizations promoting AI while hinting at team reductions shouldn't expect enthusiastic cooperation. Clarity about:

• Pilot objectives
• Staffing plans
• Managerial roles in new processes
• Post-efficiency expectations

...proves more effective than vague assurances. Excluding managers from discussions affecting their teams guarantees superficial compliance.

Avoiding Progress Theater

When initiatives stall, common responses—weekly usage reports, content generation metrics, use case documentation—often divert energy toward proving compliance rather than solving problems.

Instead of demanding three use cases per department, focus on one high-impact process (e.g., lengthy proposal preparation). Map time expenditures, identify missing information patterns, and determine where AI adds value—evaluating overall task improvement rather than isolated step acceleration.

Progress reviews should diagnose specific blockers:

• Tool limitations vs unclear business rules
• Training gaps vs redundant data entry

The former requires training; the latter requires process redesign. Recognizing unsuitable tools and terminating pilots constitutes valuable learning—not failure. Penalizing discontinued projects breeds "everything's fine" reporting.

Operational leaders constantly balance competing priorities. For AI projects to advance, leadership must demonstrate commitment through:

• Staffing decisions
• Process redesign authority
• Evaluation metric adjustments

When encountering "departmental resistance," ask for:

1. Specific task identification
2. Blockage point analysis
3. Attempted solutions
4. Required decisions

Set deadlines for departmental actions and escalate cross-functional decisions. This approach proves more effective than reiterating AI's strategic importance.

References:

- ASCM & DSCI, October 14, 2026: Supply Chain AI Workforce Readiness Report briefing.

- Jacob Dominski et al., UK Institute for Fiscal Studies, April 2, 2026: Managers as gatekeepers in the age of AI.