My summer intern finished his last day with me the way most internships end, with a review of his work. It took twenty minutes. The conversation that followed took an hour.
He is a second-year college student. He asked how I found my way into product. He asked which internships were worth his time. Then he asked the question I have been answering for a month, mostly to myself: if I were his age today, where would I begin?
He is entering a field where the thing I spent my early career learning to do, turning an idea into working software, now takes a few days. That is not a threat to him. It is a change in what the years at university are for.
AI changes what students can make. It does not remove the need to understand why something matters. That difference will shape how this generation learns, builds, and leads. What follows is what I told him.
Curiosity comes before certainty
I was building things long before I knew the word product.
My older brother became an architect. He built buildings. I built computers. Both of us wanted to understand how separate parts could become something useful, and that curiosity followed me into my career. I studied products, industries, customer behavior, and organizations. I learned through good decisions and painful ones. Over time, that experience became judgment.
I told our intern that product thinking does not begin with a job title. It begins with curiosity, and it shows up when someone looks at an ordinary product and starts asking questions. Why was it designed this way? Who does it help? What else could it become? Someone at Apple asked that last question about an object millions of people already had in their ears, and AirPods became a hearing aid.
AI can help answer these questions. Students must first care enough to ask them.
Building is easier, but judgment still takes time
AI has reduced the distance between an idea and a prototype. Students can now create software, explore hardware, and test ideas without waiting for a large team. That access is real, and so is the responsibility that comes with it.
The hardest question is no longer "Can I build this?" The harder questions are "Should I build this?" and "Who will it help?"
AI can offer options. It cannot take responsibility for the outcome.
I felt this while building Revolv the week before he left. Information about one person had become three separate records. One held an email, one held a phone number, one held the meeting history. The tools had built quickly on a weak assumption, and the fundamentals of how data represents a person were what caught it. That is the pattern. AI builds fast on whatever you believed when you asked. A person still has to notice when the belief was wrong.
He starts a database course this semester and an internet architecture course after it. I told him both will look outdated next to the newest model, and that both are what let a builder see the wrong assumption before the tools build ten thousand lines on top of it.
Students still need to test assumptions, understand people, and decide when a product provides real value. AI coding tools will not tell them when they are spinning. The tools are relentlessly encouraging. Somebody has to be calm enough to disagree.
Paul Graham recently wrote about how universities should prepare founders. He argues that universities should help students become capable builders and encourage independent projects. I agree. Students learn differently when they build something because it interests them. A personal project creates ownership, and it turns failure into information instead of a final grade.
But students should not have to develop alone. They also need experienced people who will listen, question their assumptions, and remain involved after the first project fails.
The right experience changes how someone thinks
My first internship was an office assistant role, and most of it was spent at a photocopier. I had taken the job to learn how an organization ran. What I learned was how a hundred-page document ran through a machine. Nobody thought to put the intern in the room where decisions were made.
That experience shaped how I work with interns today. I want them in real conversations. I want them to see how decisions develop, and to understand how their work reaches customers, colleagues, and the larger organization.
An internship should give someone more than a company name. It should leave that person with stronger questions, better judgment, and more confidence in their ability to contribute.
On his last day, the intern said he had learned how product development works. He also said he had gained some understanding of how I think. That was the most meaningful result of the summer for me. My reward was not a completed assignment. It was knowing that his curiosity had expanded.
Reflect without punishing yourself
Near the end of our conversation, we talked about reflection, and he admitted something. There had been a week that summer when the coding agent produced thousands of lines, praised every step, and he felt unstoppable. He was spinning. Nothing that week made the product better.
I told him I had done the same thing building Revolv. The tool never says you are spinning. It says great job. Somebody has to be calm enough to disagree with it.
Some people schedule reflection as another task. I have seen calendars with "reflection" blocked for Sunday at eight. It works better whenever a quiet moment appears. Stop, and ask a few questions. Am I moving this week, or only busy? What did the last failure teach me? Which question am I avoiding? Do I still care about the problem?
Then stay passionate about the goal and be gentle about the miss. Those are different muscles. You can be hard on yourself forever and never get there. Passion is rarely the missing piece. Steadiness is.
The conversation I am bringing to Carnegie Mellon
I leave for Carnegie Mellon University this week as part of my innovation work. I will speak with professors, students, founders, and investors, and I want to understand how each group sees this period.
I am carrying five things from that last conversation with my intern, the five I would tell any student starting today:
- Learn the fundamentals that help you understand your tools.
- Build projects that interest you, even when they look small.
- Use AI to explore, but keep responsibility for every decision.
- Find people who will support your development beyond one program.
- Reflect often, and stay passionate without being hard on yourself.
I also want to explore what becomes possible when different generations build together. Experienced leaders bring knowledge, context, and relationships. Students and recent graduates bring new perspectives, technical fluency, and fewer assumptions about what technology must become. Neither group holds the complete answer. Together they hold most of it.
That belief sits at the center of what Revolv is built on. Experience and emerging talent should be able to find each other, and stay found. An introduction is a moment. A mentorship program has an end date. What we are building is the relationship in between: the one that grows through shared work, earns trust, and still knows your story a year later.
When our internship ended, I told the student that our conversations did not need to end. He could contact me about a difficult class. He could show me a project. We could discuss his next decision. One day, I hope he provides that same support to someone else. That is how a strong network grows.
If you are a professor, a student, a founder, an investor, or an operator who has done the work and wants to pass it on, there is a seat for you at Agora. The next generation is ready to build. The question is whether we will be in the room when they do.
