What AI Level Are You Leading At?
The Question Every Executive Should Be Asking
Most leaders believe they are using AI. They are not wrong. They are also not doing what they think they are doing. Asking ChatGPT to punch up a memo is not AI-enabled leadership any more than using a calculator is quantitative finance. The tool is present. The capability is not.
There is a useful parallel in how engineers talk about AI-assisted coding. Steve Yegge mapped out eight stages of AI adoption for developers, from tentative code completions all the way to running fleets of parallel agents in fully automated pipelines. The core insight is that each stage is not just a different level of tool use. It is a different relationship with the work itself. The same logic applies at the executive level, and the stakes are considerably higher.
Leadership at its best is not execution. It is pattern recognition at speed, network activation, risk arbitrage, and the relentless work of keeping an organization pointed in one direction when drift is the default. Those things do not get automated. They get amplified, but only if you restructure around the amplification. Slapping AI on the org chart you already have produces Level 1 leaders running Level 1 organizations. Faster, maybe. Fundamentally different, no.
Here is how to think about where you are and where you need to go.
Level 1: AI as Advisor
At Level 1, AI is a conversation partner. A smart one, occasionally useful, largely contained. You use it to draft things, test arguments, summarize documents, and occasionally ask it questions you would have otherwise googled or delegated. The core structure of how you work has not changed. Meetings still run the same way. Decisions still move on the same timeline. The staff still filters information before it reaches you, and the OODA loop still turns on the speed of your calendar.
This is where most senior leaders are today. It is a reasonable starting point and a dangerous resting place. The risk is not that Level 1 leaders look ineffective. The risk is that they look fine while the floor shifts underneath them. The executives building Level 2 and Level 3 organizations are not just working faster. They are compressing decision cycles you cannot compete with using legacy structure.
Level 1 is not a failure. Staying there is.
Level 2: AI as Infrastructure
Level 2 is where the real work begins, and it is structural work, not personal productivity work. At this level, you have rebuilt the information and decision architecture of your organization around AI as a core component. You are not asking AI for opinions. You are routing signal through it, running analysis through it, and designing your staff function so that what reaches you is already processed, contextualized, and decision-ready.
This looks different in practice than it sounds in theory. Fewer status meetings because the AI-augmented reporting infrastructure surfaces what matters without a human relay race. Faster escalation because the system flags anomalies rather than waiting for someone to notice them. Tighter strategy reviews because the synthesis work that used to consume three days of analyst time takes three hours. The OODA loop does not just get faster. It gets structurally faster, meaning the improvement persists and compounds rather than depending on heroic individual effort.
The organizational design question at Level 2 is not how to add AI to what you have. It is which existing structures exist solely to compensate for limitations that AI has now eliminated. Those structures are cost without benefit. The coordination overhead that three layers of management was solving for is solvable differently. The information latency that weekly reviews were addressing can be addressed in real time. Level 2 leaders ask those questions and follow the answers somewhere uncomfortable.
Teams at Level 2 are smaller, faster, and more accountable. Collaboration increases because the friction in collaboration decreases. Decision rights get pushed further out because the information quality at the edges improves. The leader’s job shifts from managing information flow to setting the context in which good decisions happen.
Level 3: The Transcendent Organization
Level 3 is not a more advanced version of Level 2. It is a different kind of organization, and most traditional enterprises are not ready to build it. At Level 3, AI is not infrastructure supporting the organization. It is woven into the organization’s operating model so completely that the boundary between human judgment and machine processing is no longer a visible seam. The organization does not use AI to move faster. The organization’s tempo is AI-native, and everything else is calibrated to that tempo.
What this means for executive leadership is significant. The work that consumed the largest share of senior attention at Levels 1 and 2, information gathering, status management, coordination, alignment rituals, is largely automated or delegated to systems that do not require human intervention. What remains is exactly what experienced executives are actually worth: judgment on genuinely ambiguous decisions, network activation at critical moments, culture setting, risk assessment that requires contextual wisdom that cannot be trained into a model, and the forward-looking pattern recognition that comes from decades of hard-won experience.
Level 3 organizations are decoupled by design. Teams can move without waiting for permission because the decision support infrastructure makes local decisions safer and faster. The executive is not a bottleneck. The executive is a strategic resource deployed on the questions that actually require them. The org chart shrinks. The output does not.
Getting there requires a willingness to question structures that feel permanent because they have been present for a long time. The approval chains, the coordination meetings, the reporting layers, these were engineering solutions to specific problems. Some of those problems still exist. Many do not. Level 3 leaders are honest about which is which.
The Real Risk
The executives who will be left behind in this era are not the ones who resist AI on principle. They are the ones who adopt AI superficially, feel satisfied with the productivity uptick, and never ask the harder structural questions. A Level 1 leader running a Level 1 organization with better tools is still a Level 1 leader. The organization has not changed. The competitive exposure has.
Your acquired wisdom, your pattern recognition, your network, your hard-fought understanding of how enterprises actually work under pressure: none of that is obsolete. It is, in fact, more valuable when you are not burying it under coordination overhead and information management. The path to Level 3 is not about surrendering leadership to machines. It is about deploying your actual judgment on the things that actually require it, and building the organizational architecture that makes that possible.
The question is not whether AI will change how organizations are led. It already has. The question is whether you are building the structure to take advantage of it, or waiting to see how it turns out.