Annie Tsai

There’s something interesting happening in enterprise AI right now. Companies are paying for training programs, rolling out tools, making mandates stick — and then, a few months later, they’re rehiring many of the roles they tried to automate.

The AI worked, but the problem was that they couldn’t actually use it the way they’d originally planned.

It’s an interesting reversal. You don’t often see billion-dollar bets get walked back as publicly as some of these have. And once you notice it, you see it everywhere: the company that trained 80% of its team on a new AI platform, then discovered nobody knew how to actually integrate it into their workflow. The organization that spun up a whole automation initiative, hit friction when things got messy, and decided humans were actually faster. The team that got the tool, got the training, and still couldn’t execute under real conditions.

What’s happening at its core is a failure of honest assessment.

The Contradiction in the Data

Here’s what the research is actually saying, if you listen to what’s between the headlines:

Randstad found that 63% of businesses invested in AI training in the past year. That’s a massive number, and yet outcomes are flat. Adoption is growing to where about 10% of businesses are actually using AI in production now, up from 5% a year ago. But, the training spend isn’t translating to capability the way organizations expected.

Gartner warns that 50% of enterprises will lose their top AI talent by 2027 without a people-centric strategy. Think about that — attrition is expected because so many organizations simply aren’t structured to keep the people who understand how to use AI effectively.

These two signals together lead me to believe that this age old challenge continues to persist within companies: leaders are confusing activity with readiness.

They’re measuring whether training happened, whether tools got deployed, and whether mandates effectively drove activity and they’re calling it “AI adoption.” But adoption is not the same as usage. A team can complete training and still not know how to use the tool under pressure. A company can deploy an AI solution and still not have the judgment infrastructure to deploy it well. You can have all the pieces and still not be able to execute.

That’s the gap — and most organizations discover it only after they’ve spent the money, trained the team, and have to rehire to execute what the tool was supposed to do.

Readiness Is Harder Than You Think

Effort and activity are the most common measures of output — things like hours trained, budgets spent, tools adopted, and initiatives launched because these are easy to count.

Readiness is harder because it asks a fundamentally different question: When things get complicated, when judgment is required, when the AI output is ambiguous, can our organization move forward with confidence? In the real world, this is the question that actually determines whether you execute or fail.

Most organizations aren’t even asking that question, and it shows.

You see it in the rehiring. You see it in the Gartner stat — if you lose your top talent, it’s often because they’re the ones who could actually execute, and they’re frustrated that the organization won’t let them. You see it in the flat outcomes: companies are checking boxes, not building capability.

This echoes something deeper that we’ve talked about before. Back in Issue #21, we dug into the gap between mandate and skill — the assumption that if you tell people to use AI, they will, even if they don’t actually know how. And in Issue #29, we looked at what happens when organizations over-rely on AI to make decisions without the human judgment to catch what the AI misses. This is the same thread, but from a different angle. This time, it’s about measuring the gap instead of just living inside it.

Because you can’t close a gap you haven’t acknowledged, and most organizations have yet to take this step.

The Competitive Variable That Actually Matters

The organizations that are actually moving forward separate themselves with honest readiness, not speed of adoption.

The companies that walked back their automation? They eventually figured out where they actually were. Maybe it took a failed initiative. Maybe it took losing someone who could’ve made it work, forcing the organization to do an about-face and understand why attrition is occurring. And once they did, they could build from there.

The companies that are winning? They’re not racing to deploy everything under the sun as fast as it’s hitting the market. Instead, they’re taking the time to understand the existing skillset within their organization and asking: What’s our real capability with this technology? What would we need to do differently to actually use it well? What’s the gap between our current process and capacity and what we need moving forward? These are the uncomfortable questions that separate execution from the appearance of progress.

Readiness determines whether you move forward or you end up rehiring someone to do the work the tool was supposed to do.

How to Assess Readiness

Most organizations don’t have a framework to measure readiness, so they default to measuring effort and outcomes. It’s easier, and it feels like progress without requiring a deeper investment into how the organization operates at its core.

But there’s another way. And it starts with starting smaller and asking: What does readiness actually look like for us, in this specific context, with this specific tool?

That’s what I built the AI Readiness Assessment quiz to do: give you a clear picture of where you actually are across the dimensions that matter both as an individual contributor and as a leader. How’s your team’s foundational knowledge? How’s your judgment infrastructure? Your process maturity? The way you’re thinking about change management? Your culture’s relationship with AI?

Once you know where you are, you can build from there. You stop measuring effort and you start building capability. We’ll talk about that more in the next issue.

Within the context of AI, I look at readiness along seven dimensions:

• Cognitive Load Management
• Judgment vs Delegation
• Tool Fluency
• Adaptive Learning
• Privacy & Security
• Team/Collaborative Readiness
• Organizational Context Awareness

How to Measure Actual Readiness?

I’m glad you asked. As you may know, I work at Interact, and we recently launched a feature that allows me to use our tool and build exactly what you need to understand where you currently are on the scale of AI Readiness. So I built it for you to use, for free. And no, you don’t need to give me your email to take it.

https://quiz.tryinteract.com/#/6a209e40389ae822a656f130

Remember, your score isn’t a verdict. Most organizations aren’t highly ready, and that’s exactly the point. The goal for you and your team is clarity: knowing what you’re working with right now so you can stop confusing effort with execution. Once you know where you stand, you can build from there.

Next week: We’re digging into what the assessment reveals and why the gaps people find are usually where the biggest leverage lives.