Most companies assume their AI rollout stalled because they chose the wrong tool, or didn't spend enough. That has not been my experience. Every time I have watched a digital transformation lose momentum inside an organization, the cause traced back to the same two things: people didn't understand what they were being asked to do, and they hadn't yet learned to trust the tool or the leader asking them to use it.
I spent years at Udacity working through this exact problem alongside some of the world's largest technology and non-technology companies, and the pattern held regardless of industry.
Successful AI adoption comes down to two variables: education and trust. Not funding.
Education Comes Before Comfort
The first barrier is elementary. Do people understand what artificial intelligence actually covers? We tend to talk about AI as a single category, when it spans machine learning, computer vision, and large language models, among other disciplines. Most employees haven't had the time or exposure to build fluency in any of it, let alone all of it.
My answer has always been the same: invest in training.
Structured courses, boot camps, whatever gets people meaningful hands-on time with the tools. I often think about an exchange between a CFO and a CEO. The CEO says the company needs to train its people. The CFO objects on cost. It's the wrong question entirely. The real expense is what you pay, in stalled progress and quiet resistance, for leaving people uneducated.
Trust Is Built by Letting People Use the Tool Themselves
Education alone doesn't close the gap. The second lever is trust, and I built it the same way I wanted my own people to build it: by giving them direct, hands-on access to the tool within clear guardrails.
We rolled out Claude to roughly ninety leaders across the company early on, with explicit limits on what they could use it for, customer data being the clearest example of what was off the table. People applied it to account plans, communication strategy, and automating pieces of their own workload. The outcome wasn't only technical familiarity. It changed how people felt about the technology itself.
That distinction matters. I've watched the same fear surface every time something unfamiliar gets introduced: this will cost me my job, or if I get this wrong, it will cost the company. A friend of mine, by any measure a deeply successful person, told me he wakes up roughly once a year in a cold sweat, convinced he'll end up homeless under a bridge. The fear is irrational. It's also entirely real, and it occupies the same emotional register as the resistance people show toward new technology. Fear doesn't yield to logic. It yields to trust.
Give people supervised access, and that fear tends to convert into something closer to appetite. I've watched people inside my own company start pushing me to move faster once they gained that comfort, not the other way around.
Leadership's Job Is to Name the Intent, Not Just the Metrics
I think about this in military terms: commander's intent.
Every battle plan states its intent at the top, separate from the specific metrics beneath it, because the plan itself is guaranteed to break down somewhere along the way. I apply the same discipline to our annual business plans.
When leadership communicates only the numbers, people fill in the missing rationale on their own. If my team hears nothing but efficiency targets, they'll assume the goal is cost cutting and design solutions around that assumption, even when the real intent is growth. I repeat one line inside my own company more than I probably need to: we can't save our way into the Fortune 500. Everything we build is meant to increase throughput and quality, not shrink headcount. I have to restate that intent far more often than feels necessary, because a single explanation rarely travels intact through an entire organization.
Failure tolerance matters, too. Edison ran through twenty-five hundred failed attempts before the light bulb worked. A leader who treats an early rollout stumble as a crisis teaches the organization to stop attempting. A leader who treats it as one confirmed dead end keeps the organization moving.
What This Means if You're the One Rolling Out AI
None of this requires a larger budget. It requires a leader willing to articulate the real reasoning behind the change, repeatedly, and a structure that lets people build genuine comfort with the tool before being asked to rely on it. In my experience, the return on that investment shows up as speed. Teams that trust the tool and understand the intent behind it move faster than teams that are simply given instructions.
I see this same dynamic play out constantly in supply chain and transportation spend management, where persuading finance and logistics teams to trust AI-driven output enough to act on it is half the challenge.
Watch the Full Conversation
In the full discussion, I cover a few other angles, including which skills matter most as AI reshapes how businesses operate. Watch the complete conversation today.
