Back to Results

The Hardest Part of AI Transformation Isn’t the Technology. It’s the Human System Around It.

By David Hassell, Co-founder and Vice Chairman of 15Five

One of the easiest parts of an AI transformation is producing an impressive demo.

A team can show a model summarizing documents, generating analysis, or completing a task in a fraction of the usual time. Executives see the potential. A pilot is approved. Licenses are purchased. A launch is announced.

Then the real work begins.

Employees return to familiar workflows. Managers don’t know when to trust the output. People worry that experimentation will expose what they don’t know. No one is certain who owns an AI-assisted decision.

Activity increases. Business value does not.

The hardest part of AI transformation is not the technology. It is the human system around it: behavior, workflow, skills, incentives, trust, identity, and leadership.

That does not mean people are the problem. People are often responding rationally to an environment that has not given them clarity, permission, or support.

Access is not transformation

Organizations can report thousands of prompts, high license activation, and dozens of pilots while changing little about how value is created. Those measures show that tools are being used, not whether they’re driving actual change.

Transformation happens when the way an organization thinks, learns, decides, and creates value changes. That requires redesigned workflows, explicit decision rights, new management practices, and a shared understanding of where human judgment remains essential.

McKinsey’s work on scaling AI makes the same distinction. Organizations capturing more value are not merely adding tools to existing processes; they are reworking operating models, developing capabilities, and treating adoption as a change-leadership challenge.

Canva discovered a permission problem

Canva offers a vivid example. For its second AI Discovery Week, the company paused day-to-day work for more than 5,300 employees across every function and time zone. Thousands participated in 64 workshops, showcases, deep dives, and roadshows before a two-day company-wide hackathon generated 467 ideas.

Canva created protected time, offered technical and nontechnical learning tracks, brought in outside experts, and gave individual teams opportunities to explore what AI meant for their specific work. The company made experimentation part of the job rather than something employees were expected to squeeze between meetings.

The results were measurable. By the end of the week, 89% of employees rated their confidence using AI at four or five out of five, up from 72% at the beginning. Daily use of Canva’s AI tools increased by 110%.

Access to AI tools alone does not change behavior. People also need time, relevant examples, practical support, and unmistakable permission to learn.

Managers are the transmission layer

AI strategy may be set at the top, but transformation happens inside the everyday work managers lead. Theyare closest to the practical questions:

Which parts of a workflow should AI handle?
Where is human judgment essential? How will the team evaluate quality?
What should people do with the capacity AI gives back?
Who remains accountable when an AI-assisted decision causes harm?

Yet many organizations are asking managers to lead the most consequential redesign of work in decades without giving them adequate training, context, or support.

AI makes the technology more capable. It does not make change management disappear.

Managers need enough fluency to redesign workflows, coach people through uncertainty, set expectations, and keep accountability human. AI can help them identify patterns and prepare for important conversations, but managers still have to translate those insights into clarity, feedback, and action.

Build the conditions to create value

Leaders can make AI transformation more real by focusing on five conditions:

Start with a meaningful business problem. “Use AI more” is not a strategy. Choose an outcome where better speed, quality, insight, or service matters.
Redesign the workflow with the people who do the work. They understand the exceptions, friction, and judgment points that a process map misses.
Give teams protected space to learn. Experimentation squeezed between meetings will rarely produce deep behavior change.
Equip and hold managers accountable. Managers need practical training, decision frameworks, and support – and their own performance expectations should reflect the change they are expected to lead.
Measure capability and outcomes, not just usage. Track cycle time, quality, customer impact, learning speed, decision quality, error rates, and the amount of human capacity redirected to higher-value work.

HR belongs at the center

AI transformation has the potential to change roles, skills, expectations, career paths, performance standards, and trust. HR should therefore be a strategic operating partner, not a communications function invited in after the technology decision.

Technology teams can make AI available. It takes the whole leadership system to make it valuable.

The strongest transformations will feel less like software rollouts and more like organizations learning how to work differently. Technology creates capability. People convert it into capacity. Leadership determines whether that capacity becomes fear, marginal efficiency, or meaningful new value.

About the author
David Hassell is the co-founder and Vice Chairman of 15Five, an AI-powered performance management platform. For more than 15 years, he has worked with organizations to help managers become more effective and unlock the potential of their people. Connect with David on LinkedIn.