Why We Need to Protect Understanding in the Age of AI

In 1985, I helped start a division at a bank to lend to and invest in companies being acquired through leveraged buyouts (what the world would eventually call private equity). Back then, the industry was still brand new. There weren’t proven methods, templates, or experienced colleagues sitting nearby who could tell me exactly how the work should be done.

My boss was a visionary in the truest sense. A big picture, wing-it kind of guy. He could see the opportunity, and he had a strong sense of where the market was going. But he wasn't especially interested in designing the operating architecture required to get there.

So, at 25, I found myself building almost everything I needed to start and run this new business (which eventually became a large and very profitable one). The models, the approval process, the investment frameworks, the memos that made the case for why a bank should put real money behind a transaction. I even had to buy my own computer because the bank didn't provide them.

I didn't fully appreciate it at the time, but from that experience, I learned what it takes to start something with purpose and, perhaps more importantly, how deeply you need to understand something to be able to defend your point of view. In hindsight, it was my first startup/founder experience.

There’s a special kind of pressure that comes from knowing you can’t half-ass a build or, in my case, an investment recommendation that would determine whether I would earn the right to keep going. Every company we reviewed had to be understood from multiple angles: cash flow, the balance sheet, the leadership team, industry position, debt capacity, downside scenarios, and the basic logic of the deal. I had to do the research, build the projections, write the memo, and sit across from the bank’s Chief Credit Officer, a person who had seen enough bad loans to know when someone was relying on borrowed confidence.

That process taught me something I still deeply believe: Polished output isn't the same as real understanding. That idea has become even more important in the age of AI. As founders/CEOs, we need to protect the kind of work that builds understanding, especially now that AI makes polished output so much easier to produce.

Understanding Starts With Explanation

Those early investment memos taught me that you don’t really understand something until you can explain it clearly. Yes, the memos communicated a recommendation. But more than that, they exposed the quality of the thinking behind the recommendation.

A phrase like “This is a strong company” wasn’t good enough. What made the company strong? Was it the recurring revenue, limited customer concentration, gross margin, leadership experience, market position, durability of demand, or some combination of those things? And just as important, what could put each of those at risk? What assumptions had to be true for the deal to work?

Writing forced the thinking into the open. I learned that there's a big difference between reading about a business and understanding a business. You can review the materials, look at the numbers, listen to a smart discussion, and feel like you've got it. Then you sit down to write the memo, and something inside you says, "You're not done yet." Over time, I came to deeply trust that instinct.

Leadership requires the same discipline. A recommendation may sound reasonable at first, but the real test comes when you start asking what has to be true for it to work. That's when you find out whether someone has real understanding or just an answer that sounds good.

That's the point of forcing important recommendations to be written. Not to add steps, and not to slow good people down, but to see whether the logic is clear, the assumptions are sound, and the understanding is really there.

Writing doesn’t just capture our thinking. It makes the thinking better. It helps us organize what we think we know, see the gaps in our logic, and turn information into recommendations others can actually understand and use.

The world as we have created it is a process of our thinking. It cannot be changed without changing our thinking.

Albert Einstein

Familiarity Isn’t the Same as Understanding

The brain is very good at mistaking familiarity for understanding. When we hear the same idea a few times, or see a concept repeated across meetings and memos, we feel like we know it. The language, storyline, and patterns are all familiar. But familiarity can create a false sense of mastery.

This shows up all the time in growing companies. A team member may be able to talk about a customer segment, a product issue, or a market opportunity in broad terms. But when asked to explain the underlying mechanics, the tradeoffs, and the assumptions, they discover they haven't fully worked through it yet.

That's human. It's also one of the reasons founders need to build systems and disciplines that require people to demonstrate their own thinking. The best people are curious, capable learners who can dive into the information, make sense of it, and turn what they’ve learned into clear recommendations.

That's why I often insist on the discipline of writing things down. It helps people move from "I think I understand this" to "I can explain it, defend it, and effectively articulate the cost/benefits."

Writing is one of the best ways to get beneath familiarity. Asking questions is another. This is the discipline behind first-principles thinking, and the brilliance behind Elon Musk's relentless focus on it. Instead of stopping at what sounds familiar or resembles something you’ve seen before, you keep peeling back the logic until you reach something you actually know to be true. A few questions do most of that work:

• Why are we doing this?

• Why are you recommending this?

• Why is this issue and/or recommendation in this room, at this moment?

None of these questions are hostile. They're clarifying and directional. They move a conversation from the surface of an idea down to the load-bearing logic underneath it, and they quickly reveal the difference between someone who has reasoned a problem up from the ground and someone who's simply repeating a well-rehearsed summary.

We don't want to create a culture where everyone is grilled all the time. That gets old quickly and becomes counterproductive. We should be creating cultures where people understand that clarity is part of the work.

When someone owns a recommendation, they need to own more than the answer. They need to own the reasoning. That's how judgment gets built. And over time, that discipline creates better leaders.

AI Can’t Replace Real Understanding

I deeply believe every founder/CEO should leverage AI broadly every day.

AI is extraordinary. Used well, it can help teams summarize information, organize ideas, pressure-test arguments, draft options, and move faster on work that used to take far longer.

I also think every one of our leaders should be using it daily to ask questions, look for suggestions, and explore options. In this New Age of Work, I truly believe the companies that aren't leveraging AI will put themselves at an unrecoverable disadvantage over the near to medium term.

If you're open to this perspective on AI, then one of the biggest questions you need to ask is this: How should we use it?

When an analyst uses AI to help create a memo, that can be useful. When AI helps that same analyst see a missing angle or compare scenarios, that can be useful too. But if they use AI to produce the memo without wrestling with the logic, then something essential will be lost: their ability to critically think.

This distinction is crucial. A company can become faster at producing plans, summaries, and recommendations while becoming less capable of defending the thinking behind them. From a distance, everything may look more productive. Under pressure, the gaps become much easier to see.

At Ninety, we built Maz, an AI-powered companion embedded into our SaaS platform. We're building it just as much for our own needs as our clients'. Maz doesn’t exist to remove our team members from the work. It helps us organize what we're seeing, ask better questions, explore what's working and what's not, make better recommendations, and get to clearer next steps faster. Yes, Maz enhances our capabilities, but we're very clear that our people own the reasoning, the assumptions, and the recommendations.

We’ve invested millions into Maz because we deeply believe AI can help our team and every one of our clients operate more efficiently, compete more effectively, and accelerate enterprise value creation. That said, we also deeply appreciate that it can't replace human judgment, earned understanding, or the developmental value of working through logic.

What makes Maz so valuable is what it’s becoming: a powerful processor of information that helps teams complete the critical tasks that matter most to their work and their business, while also helping them learn faster. It creates more space for the uniquely human parts of the work, like judgment, meaning, trust, and care. Those are the things businesses can’t live without.

How to Build Teams That Think Deeply

As founders/CEOs, we shape the standards of our company. The way we review work, ask questions, make decisions, leverage tools, and respond to uncertainty teaches everyone else what kind of thinking is expected.

Here are a few practical ways you can protect deep thinking and develop an organization that leverages AI:

1. Talk openly about AI: Ask your colleagues what they’re experiencing. What’s helping? What feels challenging? What are they concerned about? Teams learn to use AI well when they can talk honestly about both the opportunities and the fears. There’s real emotion behind whether people choose to use AI, and most companies don’t spend enough time understanding it.
2. Ask people to note where AI helped: Have team members flag when they've used AI to get to an answer. It keeps the work transparent and shows where the tools are actually adding value, whether that's to write a memo or create a presentation.

3. Ask for the reasoning before reviewing the answer: Before responding to a recommendation from a team member, ask them to walk through the logic. What are they assuming? What options did they reject? What would make the recommendation wrong? This helps separate false confidence from true understanding.

4. Use written memos for meaningful decisions: A memo doesn’t need to be long. In fact, shorter often forces better thinking. Ask for the decision, context, assumptions, risks, alternatives, and recommendation. The goal isn’t more paperwork. It’s visible thinking: something people can see, question, and understand together. AI should help us accelerate understanding, not skip over it. The better the thinking, the faster we can move with confidence.

5. Make AI use transparent and owned: It's fine for team members to use AI to organize or challenge their thinking. In fact, it's encouraged. But they need to own the final logic. A simple standard helps: "Use the tool, then be ready to defend the work as your own."

6. Reward people for saying what they don’t understand yet: “I haven’t worked that through yet” should be treated as a responsible statement when it’s paired with a plan to learn. People grow faster when they don’t have to pretend every answer is already complete.

7. Turn review conversations into development conversations: Don’t just approve, reject, or revise. Ask questions that help the person improve how they think next time. This builds leaders who can carry more weight with greater clarity.

I'm constantly aware of the tension between efficiency and effectiveness. Both matter. We don't need to overwork every decision. But for the decisions that truly shape the company, people should know that deep thinking is not just expected but essential. That's when the work gets better, and so do the people doing it.

Better Tools, Better Judgment

Looking back, I learned to think clearly in those early days because the situation required it. There was no real way around it. I felt responsible for creating a discipline, intentionally or not, that required not just me, but the entire division, to think, write, research, defend, and learn.

Creating standards is one of the greatest responsibilities of building a company. We don’t just produce products, serve customers, and generate value. We also shape people. The people who pass through our companies carry our standards with them. They become better colleagues, better leaders, better spouses, better parents, and better citizens of whatever comes next. That’s a big deal, especially now.

It’s an extraordinary time to be alive, and especially to be a founder/CEO. But this new age also raises the bar. The best companies won’t simply be the ones that produce more. They’ll be the ones led by people who can think clearly, explain their logic, defend their recommendations, and know the difference between an answer that was generated and understanding that was earned. That’s how better tools become better judgment.