The Real AI Strategy Is Personal

Why the Shift From AI Augmentation to Systems Thinking Begins With the CEO

A few months ago, I was sitting in a board meeting at a growth-stage SaaS company when one of the directors asked about the company’s AI strategy. . . again. From what other CEOs tell me, this has been the recurring question in nearly every board meeting over the past 18 to 24 months. It is no longer a thought experiment; it is an expectation. Many VCs are shifting their entire investment thesis toward funding AI-first companies. And most are pushing their current “non-AI-first” portfolio companies to become AI-first as quickly as possible

In this case, the CEO answered well. He walked through how the company was using AI to augment people’s capabilities. For example, Engineering was adopting the latest tools to write higher-quality code faster; Customer Success was experimenting with summarization and internal knowledge assistants; and Marketing was leveraging large language models (LLMs) to produce personalized content at scale. At this meeting, I was struck by the fact that all of these activities fall in the realm of “AI augmentation.” I.e., the company was using AI to accelerate existing workflows and improve output quality. But it begs the question: “Is faster output the same thing as an AI-first strategy?”

What no one in that room asked were questions such as:

• “If AI meaningfully reduces the cost of intelligence, should we still be organizing the company around static dashboards and monthly reporting cycles?”

• “Should forecasting still be a quarterly ritual, or should it become a continuous ‘learning system’ that reconciles sales activity, product usage, and billing data in real time?

Fast forward to this past month: Mark Moses, CEO of CEO Coaching International, challenged all partners to be deliberate in helping clients think about their AI strategy. We are now using four questions to frame these conversations:

1. How is AI impacting your industry?

2. How is AI impacting your business?

3. What job can an agentic AI solution reinvent in the next six months?

4. How is AI impacting your job as CEO?

The first two questions are increasingly straightforward, though still important to clarify. McKinsey’s 2025 State of AI report shows that 65% of organizations now report regular use of generative AI in at least one function, nearly double the figure from the year prior. Adoption is broad and accelerating. Yet nearly two-thirds of respondents say their organizations have not begun scaling AI across the enterprise. In other words, experimentation is common; however, companies are still struggling with integration.

The third question, around agentic AI, is generating a lot of interest and, to be honest, a fair amount of confusion today. People are asking questions such as:

• What does agentic even mean?

• What roles can an agent play?

• How do I think about building out functions that have human+agent teammates?

According to McKinsey, roughly 62% of organizations are at least experimenting with AI agents. There is far less clarity around what to redesign and what to leave alone.

The fourth question is the one I find most interesting for CEOs to think about more deeply. At the consumer level, Pew Research reports that roughly half of U.S. adults have used a LLM (e.g., ChatGPT) at least once. But regular, structured use is far lower. Which brings us back to the boardroom. When directors ask about your AI strategy, they are partly asking about product features and cost leverage. But they are also asking something more fundamental: “Have you, as the CEO, changed how you think and behave?”

This essay is about that shift. Because the way you adopt AI as an individual will shape the ceiling of how your company adopts it.

---

My own journey through augmentation

When ChatGPT came to market, like most people, I started experimenting with it. I used it to draft emails, help with research, and summarize findings for articles. I spent time playing with prompts to make “chats” more efficient. In May of last year, I wrote an essay that summarized where I was at that point. At that time, I was intrigued by the ideas of cognition and authorship. Specifically, how do you know what is “your idea” versus what was “augmented intelligence”?

At Maxio, I was encouraging the organization to rethink workflows. At the same time, we had more than $20 billion in billing and invoicing data across our customer base. The strategic question was not whether AI could write better emails. It was whether that data, properly modeled, could reposition us as an insight engine rather than just a system of record. We wanted to be AI-first, but we also had 2000+ customers we were currently supporting and needed to keep improving the current platform, addressing tech debt, etc.

What I did not fully appreciate at the time was that my own maturity with AI was still largely at the augmentation stage. I was using it to make myself more productive. I was not yet using it to redesign how I thought about systems.

What’s the difference?

Augmentation thinking asks, “How can AI help me do this task faster or better?” Systems thinking asks, “If the cost of intelligence collapses, how should I redesign the workflow entirely?” The first improves efficiency; The second changes architecture. And this “systems thinking” is what I needed to embrace to help lead the company forward. I left Maxio in March 2025, but have continued my own AI journey.

---

A Personal AI Maturity Table

To make this concrete, I started thinking about AI adoption not just as a company journey, but as a personal one. I believed that, before you can architect AI into an organization, you have to understand how it shows up in your own cognitive workflow (i.e. how you think). That led me to sketch what I am calling a Personal AI Maturity Matrix. There are three primary dimensions that determine where you sit on the Matrix.

• Reach: How consistently are you using AI in your own work? Is it occasional experimentation, or is it part of your daily rhythm?

• Depth: How embedded is AI in your recurring cognitive workflows, such as strategy formulation, hiring decisions, capital allocation, and board preparation? Is it a peripheral tool, or is it shaping how you structure problems?

• Delegation: How much intellectual authority are you granting it? Is it editing your words, stress testing your logic, modeling scenarios, synthesizing data, or shaping first-draft frameworks that you then refine?

When you look at these three dimensions together, patterns start to emerge. I then built out five stages that reflect different ways of thinking and behaving along the AI journey. To ground this in something more than anecdote, I asked ChatGPT to synthesize adoption data from Pew, McKinsey, and the Stanford HAI AI Index to approximate how these levels might distribute across the U.S. population. To be clear, these are directional estimates, not census-level precision, but they align with what public research shows. Currently, there is high awareness, relatively lower integration, & very low systems-level redesign.

Interestingly, if you step back, the pattern roughly maps to Geoff Moore’s technology adoption curve from Crossing the Chasm. Today, most people fall into the late majority or laggard categories in their use of AI. Very few people are operating at the frontier.

Personal AI Maturity Table

• Level 0 | Avoidance
  Estimated U.S. adult population: 40 to 50 percent
  This is where most people still are. They are aware of AI. They have opinions about it. They may even feel threatened by it. But they are not using it in any meaningful way. Reach is effectively zero. Depth and delegation are nonexistent. From Moore’s lens, these are the laggards, i.e., those who do not see a compelling reason to change their behavior because they don’t know how to start, don’t think there is real value, or are just afraid.

• Level 1 | Curiosity
  Estimated population: 20 to 25 percent
  These individuals have experimented with AI, trying ChatGPT or Claude once or twice. They may have asked them to draft a vacation itinerary or summarize an article. But their usage is sporadic. At this level, reach is inconsistent, and depth is shallow. Delegation is limited to simple tasks like drafting or summarizing. This group is interested, but not yet committed. In Moore’s framework, this aligns loosely with the late majority who are observing but not yet rethinking their behavior.

• Level 2 | Utility
  Estimated population: 15 to 20 percent
  At this level, people start to use AI regularly. They use it for drafting emails, preparing outlines, summarizing research, or debugging code. Reach is high at the individual level. Depth, however, remains task-based rather than workflow-based. Delegation extends to research and light analysis, but their thinking about AI and its impact on their daily lives remains largely unchanged. This is where I was in May of last year. I was faster. I was more efficient. But I had not yet redesigned how I approached problems.

• Level 3 | Integrated Workflows
  Estimated population: 5 to 8 percent
  At this level, people are embedding into recurring workflows. It is part of how they prepare for board meetings, how they frame strategic tradeoffs, how they evaluate hiring decisions, or how they structure complex writing. On a personal level, it could include how they think about training and nutrition on a daily basis. Reach is consistent. Depth increases because AI is now shaping recurring processes rather than isolated tasks. Delegation expands to include scenario modeling, counterargument generation, and synthesis across multiple inputs. This is where I believe I sit today.

• Level 4 | Systems Thinker
  Estimated population: 1 to 3 percent
  This is the pivot. At Level 4, AI begins to influence how you design systems, not just how you execute tasks. You start asking: if the “cost” of intelligence has collapsed, what workflows should I redesign? Delegation is bounded but strategic. You define explicit roles for AI in planning, review, and analysis. This is where augmentation turns into architecture. This is where I am intentionally trying to move over the next several months.

• Level 5 | Orchestrator
  Estimated population: <1 percent
  At this level, an individual intentionally designs systems that integrate AI across domains, with governance, metrics, and defined human override. AI is not just a tool; it is an integrated layer in how value is created. Ownership and accountability are explicit. This is rare territory today, but it will not remain so for long.

There are three macro observations from this exercise:

1. If these estimates feel skewed toward the lower levels, that is consistent with the research. McKinsey data shows widespread experimentation, but far fewer organizations report material workflow redesign or measurable business impact. Personal maturity mirrors that pattern.

2. There is no shame in being at Level 1 or Level 2. That is where most people are. This technology cycle is still early, and experimentation is rational.

3. The only way to move up the table is through practice. AI maturity is not theoretical. It is behavioral. You have to use it, reflect on it, and intentionally redesign your own workflows before you can credibly redesign your company’s.

And that last point is where this stops being an intellectual exercise and starts becoming a leadership issue. Because in a growth-stage SaaS company, the CEO’s cognitive ceiling often becomes the company’s strategic ceiling.

---

The Impact of the CEO’s Personal Adoption of AI

What has become increasingly clear to me is that for CEOs in particular, learning how to use AI well is no longer optional. Not because it is trendy, but because it is becoming foundational to how decisions are made and systems are designed.

If you, as the CEO, are operating primarily at Level 2, your company will likely adopt AI at Level 2. You will encourage productivity gains. You will sponsor pilots. You will celebrate efficiency wins. But you will unconsciously limit the scope of what the company can become, because you are still thinking in terms of augmentation.

If, on the other hand, you are operating at Level 4, you begin asking a different class of questions. Instead of asking, “How can we use AI to do this task better?” you begin asking, “If the cost of intelligence has collapsed, how should we redesign this workflow entirely?” You begin asking architectural questions such as:

• Revenue Model: If AI improves our ability to predict customer behavior, should our pricing, packaging, or expansion model evolve to reflect that new insight advantage?

• Capital Allocation: If intelligence becomes cheaper and faster, how should that change where we deploy human capital versus automation across product, go-to-market, and customer success?

• Org Design: If AI meaningfully reduces the cost of analysis and coordination, which roles in our organization are still designed for information routing rather than decision-making?

That shift in your personal experience allows you to reframe both the threat and the opportunity that AI presents for how your company competes and how you build it.

---

Where do you sit?

Do I expect CEOs to print out the Personal AI Maturity Table and bring it into their next board meeting? No.

That is not the point. The point of this essay is to encourage folks to ask: “what is my actual relationship to AI?" and “how can I get better?” If you want to diagnose yourself, do not ask how often you use it as “frequency” is not the indicator of success. The real indicator of success is if your thinking has changed. So ask yourself, based on your use of AI today:

• Has your decision-making speed meaningfully improved?

• Are you seeing second-order consequences more clearly because you are modeling scenarios and stress-testing assumptions more rigorously?

• Have you redesigned at least one major leadership workflow because AI changed the economics of intelligence?

• Are you spending more time on architecture and less time producing artifacts?

The answers to these questions can help reveal whether you are still augmenting output or have begun to operate at the systems level. In my own journey over the past ten months, that has been the real inflection. I interact with AI constantly, but the meaningful shift was not volume. It was perspective. I am now using it to structure problems differently, to surface blind spots earlier, and to redesign workflows I once accepted as fixed.

The unusual opportunity in the AI tech revolution wave is that the barrier to raising your own maturity is low. You do not need a transformation budget to evolve from augmentation to systems thinking. You need to be intentional in your practice and disciplined in your reflection. In my own journey over the past ten months, that shift has been the real inflection. As I wrote about in the previous essay, I am talking to some form of AI all day/every day. But now I am using it to literally think differently

So, when the board asks for your AI strategy, they are asking about product and margin. The more important question is whether you have one for yourself.