Hook
Ask most organizations about their AI strategy and what you actually find is a collection of activity: a few people with AI subscriptions, a pilot someone ran once, a tool a department bought, a policy document nobody reads. Activity is not a strategy. It is motion that looks like progress.
The organizations genuinely pulling ahead with AI are not the ones with the most activity. They are the ones following a path: a deliberate sequence that moves the whole organization from scattered individual experimentation to AI woven into how the work actually gets done. That path is what this lesson maps.
You have personal AI fluency from the earlier tracks. This track is a different job: leading AI across a team or a company. It begins here, with the roadmap, because you cannot lead a journey whose stages you cannot name. Once you can see the four stages and know where your organization actually stands, every later decision in this track has a place to land.
Context
Stage 1: Experimentation
The first stage is individuals trying things. People use AI on their own initiative, unevenly, with no coordination. A few become genuinely capable; most dabble; some ignore it entirely.
Experimentation is necessary and almost every organization is at least here. But it is not adoption, and its defining risk is being mistaken for adoption. Leaders see enthusiastic individuals and conclude the organization is "doing AI." It is not. It has a handful of capable people and a long tail of nothing. The job of leadership at this stage is to recognize experimentation for what it is, a starting point, and to move deliberately to the next stage rather than mistaking the energy for arrival.
Stage 2: Pilots
The second stage is the first deliberate, scoped, measured use of AI on a real piece of work. A pilot is not someone experimenting. It is a chosen task, a defined approach, a stated measure of success and a fixed window to evaluate it.
Pilots matter because they convert opinion into evidence. Experimentation produces anecdotes; a pilot produces a result you can point to. The leadership job here is selection and rigor: choosing pilots that are visible enough to matter and contained enough to succeed, and insisting that each one has a real measure of success defined before it begins, not invented afterward to justify it.
Stage 3: Scaled adoption
The third stage is taking what a pilot proved and extending it across the team, the function or the company. The successful approach becomes the normal way that kind of work is done, by many people, not a few.
This is the stage where most organizations stall, and the gap between Stage 2 and Stage 3 is the single most important feature of the whole roadmap. A pilot succeeding is not the same as an organization changing. Scaling requires training many people, standardizing the approach, changing established habits and managing genuine resistance. It is real organizational work, and it is the subject of the next several lessons in this track. A leader who treats a successful pilot as the finish line, rather than the start of the hardest stage, will preside over a graveyard of promising pilots that never became anything.
The bottleneck is human, not technical.
Stage 4: Operating rhythm
The fourth stage is the destination: AI is simply part of how the work gets done. It is not a project, an initiative or a special effort. It is in the standard workflow, the default tooling, the way new people are onboarded, the assumptions of the operating model. Nobody calls it "the AI initiative" anymore because it is no longer separate from the work.
Reaching operating rhythm is what "AI adoption" actually means as an end state. Everything before it is transition. The leadership job here shifts from driving change to sustaining a capability: keeping the tooling current, the skills fresh and the governance sound, which the later lessons on the operating model and failure modes address directly.
Where the roadmap is really used
The roadmap's value is not as a diagram. It is as a diagnostic. Its first use is brutal honesty about where your organization actually stands, which is almost always one stage earlier than leadership believes. A company with enthusiastic individuals usually thinks it is at Stage 3 and is actually at Stage 1. Naming the true stage is the unlock, because each stage has a different leadership job, and doing Stage 2's job when you are really at Stage 1, or assuming Stage 4 when you are stuck at the Stage 2-to-3 wall, is how AI efforts fail. The rest of this track equips you for each stage. This lesson's contribution is the map and the honesty to read your own position on it.
Steps
Step 1: Diagnose your real stage, honestly
Place your organization on the four stages, and discount your own optimism. Look for evidence, not enthusiasm. Stage 1 if AI use is uncoordinated individuals. Stage 2 if there is at least one scoped, measured pilot. Stage 3 if a proven approach is genuinely being used by many, not a few. Stage 4 if AI is in the default workflow and nobody calls it an initiative. Most organizations are a stage behind where leadership assumes. Name the true one.
Step 2: Know the leadership job for that stage
Each stage asks something different of you. At Stage 1, move deliberately to real pilots rather than admiring the experimentation. At Stage 2, select and run rigorous, measured pilots. At Stage 3, do the hard organizational work of scaling: training, standardizing, managing resistance. At Stage 4, sustain the capability. Be clear which job is actually yours right now.
Step 3: Choose your next pilot with care
If you are moving into or within Stage 2, choose the next pilot deliberately. The best first pilots are visible enough that success matters and is noticed, contained enough that success is achievable, and tied to a real, recurring pain so the result is meaningful. Avoid pilots that are too sprawling to succeed or too trivial to matter.
Step 4: Define success before you begin
For any pilot, write down what success is, in measurable terms, before the pilot starts. Time saved, quality improved, capacity created: name the measure and the target up front. Success defined beforehand is evidence. Success described afterward is a story. The next lesson, on the business case and ROI, makes this concrete; the discipline starts here.
Step 5: Plan the move across the Stage 2-to-3 wall
Whatever stage you are in, look ahead to the gap between a successful pilot and genuine scaled adoption, because that is where the roadmap breaks for most organizations. Begin planning now for what scaling will actually require: training many people, standardizing the approach, changing habits, managing resistance. Treating that wall as expected work, not an unpleasant surprise, is what gets an organization over it.
Recap
- Scattered AI activity is not a strategy. Organizations that pull ahead follow a deliberate path from individual experimentation to AI woven into how work is done.
- The roadmap has four stages: Experimentation (uncoordinated individuals), Pilots (scoped, measured use on real work), Scaled adoption (a proven approach extended to many) and Operating rhythm (AI in the default workflow).
- Most organizations stall at the gap between Stage 2 and Stage 3. A successful pilot is not a changed organization; scaling is real, hard organizational work.
- The roadmap's main use is as an honest diagnostic. Organizations almost always sit one stage earlier than leadership believes. Name the true stage.
- Each stage demands a different leadership job. Diagnose your real stage, do that stage's actual job, choose pilots with care and define success in measurable terms before any pilot begins.
A roadmap needs fuel, and the fuel is a credible business case. The next lesson covers how to justify AI investment and measure its return without falling for vanity metrics.