What Does an AI-Enabled Organization Actually Look Like?

88% of organizations are using AI in at least one business function. Only 6% are capturing enterprise-level value. These figures are coming from McKinsey report from November 2025. Read those two numbers together and you'll see something that most AI conversations don't want to say out loud. The tools are not the problem. The tools exist, they are capable, most organizations have already bought them, and almost none of the organizations are getting the returns they expected. So the real question is, what is actually going wrong? Before we go further, a quick note on scope.

Throughout this series, when I say AI, I mean generative AI. Large language models. Tools like ChatGPT, Copilot, Claude, the wave of AI has been reshaping how organizations work since 2022, the moment when ChatGPT was released. There are other types of AI, of course, predictive models, computer vision, classical machine learning and others. They matter, but they are not the focus here. This series is about the gen AI and what it demands from organizations that want to do it well. Here is what most organizations do when they start with AI.

They identify a capable tool. Normally, it's Copilot, ChatGPT Enterprise, Claude or others, and they deploy it. The logic is straightforward. The tool is powerful, give people access, run some training sessions and productivity will increase. What actually happens is slightly different. Normally, a small group of people, group of employees, usually the ones who were already curious and self-directed, start using the tool intensively. They find it really valuable and leadership points out to them saying that the rollout is actually working. Meanwhile, most employees barely touch it. Not because they are resistant or unintelligent, but because nobody gave them a compelling reason to change a working habit for an unclear benefit.

And leadership ends up asking a question that nobody has a satisfying answer to. Why the impact is so limited when the tool is so capable? The most common answers are we need better change management, employees are resistant, we need to find better use cases and others. All of those might be true, but none of them is the root cause. Here is the distinction that most organizations are missing. An AI-enabled organization is not the one where AI tools are available. An AI-enabled organization is the one that has systematically redesigned how its whole work gets done.

In ways that could not be done without AI. Three words in that definition carry all the weight. Systematically redesigned means it is not just pilots. It means that organization has looked at the work that matters most, the decisions, the processes, the outputs that define what the organization actually does and asked how does it work differently when AI is reliably part of how it gets done. Core work means not the internal memo summarization, and not the meeting transcription. Those are, of course, useful, but they are not the transformation.

Core work is the analysis, the client interaction, the products, the operation, that defines the organization. And could not be done without AI. It means that change is structural. If you remove the AI tomorrow and nothing fundamental changed about how the organization operates, you have not achieved AI enablement. You have achieved just AI adoption. Let me make this concrete. For years, the standard way to present a project inside the company was PowerPoint. You collect information, structure slides, design visuals, present. And the artifact is a deck, a PowerPoint file.

Something interesting is now possible. For certain types of projects, it is actually faster and more convincing to ask an AI system to help you generate a simple interactive prototype. Something you can actually click through, then to build slides. Instead of presenting the idea, you present the working thing. So the artifact of work changes. When the artifact changes, the workflow changes. When the workflow changes, the skill that matters changes. The role of the person doing the work begins to shift. This is a small example, but it illustrates something really important.

AI doesn't only make work faster. Sometimes it changes what the work actually is. And most organizations are deploying AI as if it were a faster way to do the same thing, a better version of Excel or email. If that is your mental model, the strategy is simple. Deploy, train employees, and then hope. But it misses the transformation. So what does organization actually need to do in order to get there? The single most consistent thing I have seen missing is not tools, not budget, not talent. It is shared concrete picture of what the destination looks like.

Most companies have AI in their strategy document. They have allocated budget. They have someone responsible for adoption. But they do not have a clear specific answer to this question. What does a successfully AI-enabled version of this organization looks like in three years? Without that picture, everything becomes a pilot. You run experiments. Some of them work. Most of them don't. Nobody can tell whether any of it is adding up something meaningful to the company. Because there is no definition of meaningful to measure against. This is what I call the absence of a target state.

And it is in my experience the single most common reason that leads to AI initiatives under delivery. I want to describe a pattern which I have seen play out in many companies. Here's how it typically begins. Someone takes the lead on AI, whether an internal team or an external consultant. The first instinct is the reasonable one. Go to different parts of the company and ask what problems AI might help them to solve. The logic is very clear. You need to understand where the problems are before you will prioritize anything.

So ideas are collected, use cases are prioritized, pilots are built. A year in, the organization has a long list of AI initiatives, each looking compelling in isolation. Productivity gains exist. Efficiency improvements are also there. Some genuinely impressive demos are flying around the company. And then someone asks a basic question. What all of these things are actually adding up to? Is the organization actually becoming more capable in any meaningful durable way? Or we are just accumulating collection of useful but disconnected tools? In most cases, the honest answer is closer to the second.

The company built things that were technically impressive and organizationally irrelevant in the broader sense. Not because the work was bad. Because nobody has defined where the company was trying to go to. The use cases were solving local problems. Nobody had asked the organizational question, the broader ones. What should you do instead? Before the use cases pipeline, before the tool selection, before the training program, you should have a clear answer to the following question. What does this company look like in three years if we actually do this well?

That is the question that Target State Canvas is designed to help you to answer. It is a one-page tool that walks the leadership team through the questions. What does this organization do that AI cannot do? What does AI make possible that was not possible before? What does core work look like in three years? What stays the same? And this is actually where the real conversation begins. What are we willing to change about how we work? But before we pull out any Canvas, there is something simply worth doing first.

Which brings me to the one thing I want you to do this week. Write two sentences. Not a vision statement, not a strategy document. Two honest sentences that answer this. What does an AI-enabled version of your company actually looks like in three years? Specific, concrete, what actually changes about how the work gets done? And then ask every person in your leadership team to write two sentences. Independently, before you compare. Do not discuss it first. Just ask for two sentences each. Then look at what you have.

In my experience, the answer will be different in ways that matter. Someone will describe an organization that uses AI to move faster. Someone else will describe one where AI handles routine work and humans focus on judgment. A third one will describe AI embedded in customer interaction. And the fourth will describe internal productivity. Those are not the same organizations. If you are running an AI initiative without a shared destination, you are running multiple strategies simultaneously. And wondering why it doesn't add up. The gap between those two sentences, answers from these people, is your first AI problem.

Not a technology problem, a definition problem. And definition problems are worth solving first. Because everything else, your use case selection, your tool choices, your training program, should be shaped by where you are actually trying to go. If you are watching this video on YouTube, the Target State Canvas, the research I referenced, and everything else for this episode is on the dedicated page for this video. The link is provided in the description. The next question worth asking is why do so many AI projects stall even when the target state is clear?

And this question will answer in the next episode.