At 9am, the chief executive announces the organisation’s AI future with standard digital ambition. “We are empowering everyone with agentic intelligence,” they write in an all-staff email, “and ushering in a new horizon of productivity.” At 9.15am, an employee asks their manager a more ordinary question: “Does this mean I still have to complete the weekly report by hand?”

The manager is not the final mile of AI adoption. They are the place where a broad ambition either becomes practical—or becomes theatre. When tools arrive without redesigned processes, managers are forced to act as shock absorbers. They must keep delivery stable while translating vague mandates into actual daily routines.

01 –

Managers are not a distribution channel

When an organisation decides to introduce generative AI, the typical communication strategy is a linear cascade. Leaders draft a memo, HR designs a mandatory module, and the program office produces a prompt deck. The manager is treated simply as a link in the wire—someone whose job is to hit “forward” on emails and ensure their direct reports completed their training before Friday.

This view of management is both clinical and incorrect. A cascade of information is not a transition of work. Managers are often given broad, abstract encouragement, while their teams need clear, practical direction.

02 –

The confidence gap has a human face

When you ask CHROs if AI is integral to their future strategy, the response is near-unanimous. But when you ask if their line managers are equipped to guide their teams through this change, the consensus crumbles. This discrepancy has profound implications for adoption.

03 –

The impossible middle-manager job

Middle managers are caught in a classic operational bind. They are instructed to welcome the new tools with curiosity, while being held strictly accountable for legacy targets. They hear “innovate” from leadership, and they hear “we are overwhelmed” from their teams.

04 –

AI is a management practice, not a tool rollout

When employees feel supported by their immediate manager, they are 2.1x more likely to use AI in their weekly work. The reason is not technical; it is psychological. A supportive manager provides permission—permission to make mistakes, permission to learn slowly, and permission to stop doing work that no longer makes sense.

For non-users, the barrier is rarely mechanical. Around 44% of employees who do not use AI say they simply do not see how it can help their specific tasks; only 16% report a lack of software access. You cannot prompt your way out of a relevance problem. Only a manager can help an employee see how their specific role can change.

05 –

What good managers actually do

“Managers are not a human version of Slack.”

– Suchetana Bauri

“A webinar cannot solve a workload problem. Nor can a prompt library solve an incentive problem.”

– Suchetana Bauri

06 –

Stop training managers like employees

Most corporate training treats managers and direct reports identically: as individual operators. They sit in the same sessions, write the same practice prompts, and receive the same advice. This ignores the manager’s actual responsibility. They do not just need to use the tool; they need to manage the transition of work.

07 –

The manager enablement contract

Most corporate training treats managers and direct reports identically: as individual operators. They sit in the same sessions, write the same practice prompts, and receive the same advice. This ignores the manager’s actual responsibility. They do not just need to use the tool; they need to manage the transition of work.

08 –

From cascade to conversation

When adoption works, it looks less like a cascade and more like a conversation. Good ideas travel upwards because there is a clear route for them. Risks are identified early because managers have permission to talk about what is failing. Enabling the middle layer is not an administrative burden; it is a prerequisite for making AI genuinely useful at work.

09 –

The missing layer

If you bypass managers, adoption remains performative. Logins might rise, but the underlying work is unchanged. You produce more content, more slides, more emails—but not better outcomes.

If leaders want sustained AI adoption, they should stop giving managers messages to cascade and start giving them the authority, time, clarity and support to lead.

About Suchetana Bauri

Suchetana is an AI enablement consultant and communications strategist. She helps organisations make AI adoption understandable, practical and human-centred.