70% of the change isn't technology — it's you.
False assumptions break plans.
Here's one: "When we implement AI, the result depends on our technology."
(No.)
BCG analyzed 1,500 organizations implementing AI. Their conclusion: the success of transformation splits like this:
- 10% algorithms and technology
- 20% actual systems and infrastructure
- 70% people and culture
Let that sink in.
You could have the best AI money can buy. If your people don't change how they work, you get nothing.
Why 70% Is People
Because AI doesn't implement itself. Humans implement it. And humans are creatures of habit.
You've been doing reporting the same way for 10 years. Now the AI tells you the answer before you ask. Do you use the answer, or do you go verify it the old way first?
Your team has a process for approval. Now an AI can handle 80% of approvals automatically. Do you trust it, or do you add a human review layer?
Suddenly you're not a technology adoption problem anymore. You're a trust problem. A power problem. A "what does my job mean if the machine can do it" problem.
The 10%
The algorithm is the easy part. You buy it. Or you train it. Or you find it on GitHub.
ChatGPT, Claude, Gemini — these are commodities now. The difference between one and another matters less than you think.
What matters is what you feed it and what you do with the output.
That's not technology. That's judgment.
The 20%
Integration. APIs. Data pipelines. Clean databases. Security. Governance.
This is the work nobody talks about because it's boring. But it's the reason 60% of AI projects fail: the integration is harder than the algorithm.
Your customer data is in five different systems. None of them speak to each other. Suddenly you need to unify it for the AI to work.
That's infrastructure work. Important. Technical. But still not the main problem.
The 70%
This is the hard part.
Culture change. Decision-making rethinking. Who has authority now? What gets automated, what stays human? How do we measure success differently?
A sales team that's trained to negotiate with customers hits a wall when the AI suggests a price. Their entire identity is "I'm good at negotiation." Now a machine is doing it. What are they good at now?
HR teams built their power on gatekeeping hiring. Now an AI pre-screens resumes. Their job shifts to "train the AI to not discriminate" and "make the final call." Smaller role. Different role.
These are the fights that kill AI adoption. Not technical fights. Fights about meaning and power and identity.
The Companies That Win
The ones who address the 70% first.
They don't start with "buy the best AI." They start with "who are we becoming?"
They run pilots where people work alongside AI, not under it. They let them feel what changes. They ask what worries them. They redesign roles before the AI arrives, not after.
They make the unpopular move: promoting people whose skills are becoming redundant into new roles. Not out of kindness — because those people understand the old system deeply and can bridge to the new one.
They treat the transformation as change management, not tech adoption.
The Real Cost
People think AI transformation costs money because of the software.
Nope. The software is 10-15% of the cost. The rest is people: training, new hiring, redundancy packages, consulting, delayed decisions while you figure out the new way.
And time. Especially time. Culture doesn't change in a quarter.
What This Means
If you're leading an AI transformation, you're not really leading technology. You're leading people through disruption and rebuilding how they see their job.
That's harder than any algorithm.
It's also the only thing that actually works.
This is part of my series on the future of work. In each post I take a specific research finding, prediction, or move — and ask what it means about the organizations we live in.
If you're managing through an AI transformation, this gets into the real work: The Little Book for New Managers — it's specifically about this.
Want to read the next one?
More changes coming in how we organize work.
Sums it up, Lior
Want more on managing in the AI era?
The Little Book for New Managers has helped 30,000+ managers figure out what actually works.
Get the book ($19) →