Amit Chopra on Life After the Nine-to-Five and Careers in the Age of AI

Hi, everybody. A lot of us have the kind of jobs that take up a lot of our time, a lot of attention, and maybe even define a large part of our identity. Now, as all things go, we know that our job won't last forever. What does a fulfilling life after the nine-to-five ends even look like? To help answer this and a whole lot more, I talk with Amit Chopra, someone who I've known and followed for decades who has a very fresh and balanced perspective on this topic. So let's get started.

Amit, it is great to chat with you. I feel like I've known you forever, which makes sense because I feel like when I joined Visual Studio and DevDiv in 2007-2008, you were one of the people I worked with very closely on and off for my entire 10 years there, my first time around.

So, great to meet with you, and I want to start by asking the usual question that I always do, which is please introduce yourself and tell us how you got into this fun world of tech.

Yeah, well, likewise. It's been a fascinating journey together, working on so many interesting projects back in the Visual Studio days. I actually found one of the old t-shirts. It's not Visual Studio, but Visual Studio Rangers. This was one of the programs that got kicked off, I don't know, probably in 2010-ish or maybe even before that.

But Visual Studio, that part of my career, has probably been the most exciting journey in terms of the things I've learned, the things I've made, and the impact it has caused on the industry. So, a little bit about myself, Amit Chopra, and I'll give you a little bit of my journey before I introduce myself in more detail. I don't have a title these days, which makes it very interesting.

When I say “Amit Chopra,” I pause for a few seconds to say, okay, what do I say after that? I'm retired, I'm not retired, what am I doing? We can have a conversation on that later. But getting into tech was, I don't know, a lot of things that have happened to me seem to be very accidental. Sometimes things took me in a certain direction when I wanted to go in a different direction.

As early as the late 80s, computers were still not a big thing back in India. Ours in school, when I was in high school, was the second batch that actually had access to computers. And by computers, I mean there was a BBC microcomputer, like the keyboards, and that was pretty much the computer. We would plug it into a TV and run some version of BASIC on it.

Then a more professional machine that they got, which was only given to a small set of staff people, was a Tandy Radio Shack. It had two floppy disks and a reader for attaching a tape drive to that. Those were the first two machines I saw. And I think what fascinated me most was that those machines used to sit in an air-conditioned room in our school all the time.

This was how much computers were. You got a dedicated room, an air-conditioned room. And growing up in Delhi, it used to be hot. I'm like, I have to find a way to get inside that room. That drive itself motivated me to learn as much as I could.

There was no internet, there were no CDs, there was absolutely nothing. Anybody who grew up in Delhi would be familiar with this place called the Red Fort, the Lal Qila. Behind the Red Fort, there used to be a market that happened every Saturday. It was called the Chor Bazaar, the Thieves' Market, literally translating, for somebody who's not familiar with Hindi. They used to sell stolen stuff, pirated stuff, and you name it, you could go find it.

If your motorcycle got broken the night before, you'd probably find your parts there. That's what that market was for. And I would go there and pick up, for 25 pesos or 50 pesos, I don't know, 2 cents, 3 cents in today's terms, the PC Magazine and the Byte magazine. I used to just read them. I did not understand how these things work. I'd never actually sat down on a computer.

And interestingly, I found a very like-minded friend back in the office in my school. This was 10th and 11th grade. He was equally fascinated by computers. And for us at that point, computers actually did not mean the computers that you're seeing today. For me, computers at that point was actually fascinated by Star Trek. Mr. Spock, right? That was like a role model. He would play 3D chess.

Both me and my friend used to be quite fascinated with, oh my god, how is he doing that? What is this logic? And I have sheets, I have some pictures somewhere, where both of us used to sit and draw flowcharts for everything possible in life. We would sit and get fascinated by, okay, let's solve this using a flowchart. And that got us both quite excited into this field.

When the school actually opened up these computers for us, it was an excuse to go sit in the AC room and try something out. I think that just got me interested. But I think the real interesting stuff happened—I picked up a lot of programming at that point, just by dabbling.

We used to use a lot of Logo turtle. You had this triangle on the screen and you would say move forward 50, turn right, repeat five times. That was programming for me. For almost a year, that's all they taught us in school. There was nothing else. But during 12th, I got a little more serious into programming. I understood BASIC quite well. I had a very strong hacker's instinct.

At the same time, I was quite foolish. I broke into one of our school computers and left my name saying “Amit was here,” and got suspended. But that gave me a little bit of limelight. In 12th, that's all you need. High schooler, what else do you want? “Hey, there's somebody who's famous, who broke in.” And teachers noticed me in good and bad ways, both, right? That was the thing.

But in 12th, our teacher who used to use the Tandy Radio Shack was building something, a fee billing system. She left because there were so many opportunities happening in 89, 88. Anybody who could spell computers would probably get a job. She and her husband, who was also in tech, left the school and that project was orphaned. One of my teachers approached me and said, “Hey, would you be interested in finishing this for us?”

I was excited, I was nervous, and I said yes. It was quite nice to see that during our summer between college and school, I ended up finishing that, and the fee billing system was used for almost seven, eight years after that, running on the Tandy Radio Shack machine. That was my first professional software. But here's where the twist came.

I was actually not that interested in a career in software, because I didn't even know that careers in software existed back in 88, right? We didn't have role models. We didn't know much about tech. All I was fascinated by was to be a chef. And it was quite interesting, when you're a single male child in a family and you want to take an unconventional career in India back in 87, 88.

It's not something your families look up to. So there was a big protest in the family, and I protested that, okay, I'm not going to do engineering. That's what 18-year-olds do, they rebel. And then I finally convinced myself that I have to go and do accounting. So from this very interesting career path of doing great academics in physics, chemistry, science, and all, I switched to accounting.

For three years, that is what I did. At the same time, I didn't leave my love for computers behind. It would be like, okay, we have a college festival, and let's do a stock exchange mock-up application. So I would volunteer myself and build something. I continued to program quite a bit even though I was doing accounting and I hated it.

I realized that's actually a very powerful sweet spot for me, where I know tech, I know programming, I know accounting, and there was a massive demand for that. I started building these financial accounting systems. I got a pretty nice project in Dubai, building a seven-language multi-currency accounting system for them. And I think that just put me on this very interesting career path after that. One thing just led to another.

There was literally a story from every job that I've changed: from Microsoft to Sun Microsystems, and then PricewaterhouseCoopers, and then back to Microsoft for a very long time, then switching gears to Google, and then, where did I go? Facebook, and then to Amazon, and then a startup, Gemini, a crypto exchange of all the things, right? And then back to Microsoft after a good 30 years of a full cycle. It's been a very rewarding journey.

What I have been doing now is mostly using my time to help very early stage entrepreneurs who are making their first startups, and students who are super excited about tech. If I can help them out and guide them in any way, that's where I spend most of my time. So that's me and my journey. I think I took almost like a good 10 minutes describing 30 minutes. But now you can pick on anything and we can have a fun chat about that.

No, this is fascinating because, a common theme I've seen in many people who had a fulfilling career in tech is this unconventional route that you are describing. I've never met anyone who said that they knew from a very early age they want to get into big tech and they're going to take all the right steps to get there. I'm sure those people exist, but at least in my opinion, many of them do not have the kind of rich stories that you and others who have been in the same spot like you have had.

I think that's actually really interesting. So you mentioned that you started with Logo as your first programming language, and then you went to BASIC as a thing that you were building. Were you building things just for fun, or was there a project or some kind of a goal that you had that was given by, say, a school or a job that you're working in?

A lot of what I actually did was just ideas that I came up with, right? It was just, hey, can I actually use computers to do that? It goes back to the same thing me and my friend Bharat both did back in 87, which was to solve everything from a flowchart. We could put any decision of life, anything that came to our mind, into a flowchart.

And the same thing happened to me: can I actually code this thing? Coding was fascinating. This is something my mom tells a lot of people, including my kids and my wife: Amit would literally go into a room for hours, and nobody could disturb him. The only way she would signal something was to turn the light switch on and off, so it would get my attention. Otherwise, I was literally zoning myself in.

These were not powerful machines, PC XT. I don't know at what speed it ran, but you'd try to build something and it could take minutes. So you had to be very thorough about what you were doing. But I kept looking for these opportunities, and people gave me ideas as well. Like I said, in college, a friend of mine wanted to do something interesting for one of the college festivals.

He said, “Can we make something like a mock stock exchange?” It was quite a fascinating idea. Okay, let's figure out how this has to be done. Then somebody reached out to make a very interesting application. This was the Psion handheld computers. This was a British company, very small form factor, 2 to 4K of programmable memory, and it ran some strange version of BASIC. I don't even think it was BASIC. It was some of their own proprietary language.

He said, “Make me an application that helps me manage my accounts, but at the same time, it should not look like an accounting application.” Because he did not want to get into trouble with the local authorities, he had an interesting idea: let's code everything as phone numbers. This becomes like a telephone directory where I put certain codes for every phone number, and I know who owes me how much money.

It turned out to be a very interesting app. I still look back and say, what did I actually create there? But somebody gave me an interesting idea and we just used a machine to solve that. Two other similar ones were programming the Intel 8051 chip. I don't have a background in engineering, right? But I was always like, okay, I can figure this out. You give me enough time and a book to read.

We started programming the Intel 8051. It actually had just two kilobytes of programmable space, and everything had to be done in assembly. And I think my flowcharting skills came in extremely handy. I had to thoroughly write everything down, every register chain, walk it through by hand on paper, and make sure before I burn this EEPROM. Because if I destroy this EEPROM, I won't get it back for 24 hours.

We had to put it under ultraviolet light and actually erase it before we could do the next set of programming on it. And we ended up making two very interesting applications. One was teaching a drilling machine to automate itself. We could take a circuit board, drill it by hand, and every sequence was recorded. We moved three centimeters, we pressed the thing down, and everything was encoded in codes that we understood.

Like, okay, this means move three steps, just like what we did with Logo, and created that application. And then another one was, we had a lot of issues with heat and oil heating in transformers, and what we needed was something that would just read analog test signals. Something that an Arduino or Raspberry Pi, or even one of the most basic things, can do today.

But with 2K, it was quite an interesting challenge, being so thoughtful about every little line of assembly code that we were writing. It was three of us in the team, and we would challenge each other: okay, you've got 80 lines, I'm going to try and reduce it to 78 lines or 77 lines.

So a lot of these things gave me a lot of exposure to a broad spectrum of problem spaces, a broad spectrum of how to solve that. That's why, over my life, I don't think I ever got married to a specific technology or a solution, or said, I'm a huge domain expert on these. I've always felt technology is a great means to an end. You pick what you're most comfortable with.

Even back at Microsoft, I was creating games in Excel. I had a version of Snake that was running in Excel. Why? Because I could just do it. It was just fun for no reason, right? I mean, nobody in the right mind would use Excel to play Snake. But hey, we have VBA and we have grids and we can change colors. Let's make a game of Snake using that.

That is really cool. Isn't Excel one of those apps that had an Easter egg where it had a flight simulator? At least the earlier versions of it.

Yes. On a lot of Microsoft products I remember, from the very early days, Easter eggs were a big thing. You would discover them, and you'd go to magazines to find them out. There was no internet where you could go sneak around, so it was something I would show off at some of the demos. But I think a couple of years later, there was this big thing that said, “If a company cannot control what people are putting in, that's not enterprise-grade software.”

So even though the intention was so nice and so much fun, you would get credits, and you would get these Star Wars themes show up if you just pressed the right sequence of keys in different products. A lot of them got pulled out eventually. I don't think any product—or they're so good that they're so well hidden. But then what's the point? If you can't find an Easter egg, there's no point.

Yeah. An interesting transition as well for you is that you've been talking a lot about coding and doing things at a very deep level, optimizing memory, that you have 2K constraints to deal with. At least I feel like ever since I've been working with you at Microsoft, you were not a developer. You were more on the product side of the house. How did you make that transition, and were you happy doing that?

Oh, more than happy, actually. I'm a very lazy person, so sitting and writing code for that long period of time is just not me. Though I always keep my hands very, very dirty. Something new comes up, I'm trying that. Before ChatGPT came out, OpenAI released their library, and you could access them through the API. So I was super fascinated with what we could do. There's a post that I wrote about creating Bollywood scripts.

I created a small front end where you could put a one-liner, say, “I have an idea of a movie,” and you would put two lines. What it would do is go back and forth with ChatGPT using the OpenAI APIs, and then come back with titles, characters, names of characters, a storyline, an image. So with everything new, I make sure I understand it enough that I can see how best to use it to solve a real-life problem.

And that's where the product side comes in, right? As I said, I always felt tech is a means to an end, and product is when you start defining the end. What is the problem that the person is trying to solve? When we were part of the Visual Studio team, I think I spent almost 13 years just doing product for devices, device development.

One of the pain points really was with Windows Phone and Windows Mobile: just the diversity of the screen sizes, the form factors that were there. It was very painful. I used to write applications, took part in a lot of hackathons, and I knew it was a pain to sit down and test it out on so many different things.

And even before Apple had their Xcode add some of those features, Visual Studio 2005 had this multi-screen kind of designer experience that you could actually start designing. You could switch a button, and you would change the form factor for you. That's where my product thinking came from. Now, if I had to engineer that, no. I really had to depend on our architect, the engineering lead, to figure out how do you create a Windows Mobile environment natively inside of your desktop?

They're two very different beasts altogether. Would just virtualization work? I don't know. That's not where my strength really is. So I depend on some of those people, but I can give them ideas and say, “Hey, this is a genuine problem. If we solve for that, customers are happy. You figure out.” This is not a territory I'm going to step in unless I'm extremely comfortable talking about the nuts and bolts.

I can if I have to, but that's not where I want to kind of develop my standards. Can I spot problems? Can I suggest people? Now, the good thing is most of us have a coding assistant at our fingertips, and that's kind of blurred the line between whether I'm a product person or a coder. I'm still giving a lot of commands to my ChatGPT and some of the other tools and saying, “Hey, build this for me.”

And we can look at the efficiency. Who cares about optimization these days? Memory is cheap, storage is cheap, network is cheap. So let's just fix the problem. That's where I think the transition to product happened quite naturally to me, because I was just spotting problems. That's what I feel product managers should be really doing. Hey, can I solve this for a customer and find the best way of doing it?

Yeah. The funny thing you mentioned about the device panel in Visual Studio is that it's actually one of the first projects that I worked on at Microsoft. I think what you're describing is that Expression Blend had the device panel where you could swap between the portrait, landscape, and different views. And I think for the Mango release of Windows Phone, you had accent colors as well. You could pick and choose. And that was shared, of course, with the same code base inside the designer in Visual Studio.

So you could actually do that swapping between them. And it was convenient that I think Phone ran a version of XAML as well. So the runtime and what you saw at design time were mostly one-to-one. And then when web became a thing, was it Universal Windows apps, like UWP? That's when we had the whole web runtime running inside your environment as well. So you could see your code running in real time. Those were interesting times in terms of the technologies we invented in some ways of solving a hard problem.

Yeah, these were hard problems. Problems keep changing, right? You solve one, you discover another one, or you actually end up creating another one as part of that particular process. But for my developers, the community of people who were making a living off building Windows Phone apps, I think that was literally a blessing on how fast they could now test something. You finish your code, half the code is auto-generated by apps or snippets, or Stack Overflow used to be the source.

Okay, I get something from there and I can use it. But testing was a huge pain. I think solving for some of these problems have just changed how fast people can get things out. This is 15 years ago, and now I'm literally seeing people have 20 features being pushed out on a daily basis. Everything is fully automated. There's an agent coding for them, an agent that's reviewing for them, an agent that's pushing things into production, an agent that's documenting things.

But I personally think that frees up people's minds to focus on even harder problems, right? What other things can be solved using tech, rather than just sitting and optimizing tech, which was fun back in the days? I would love to do that, and my team used to love to do that, but now it's like, what's the next big thing we can work on?

Yeah, which is one of the things I often see you writing about. You're one of the few people that I know that was able to successfully transition from working into tech, and then a post-tech, post-career kind of a world where you're doing the things that you greatly enjoy.

So walk me through a little bit of that. How were you able to mentally be in a state where, you know what, I am done with the nine-to-five, the corporate grind, and now I'm going to do something that is entirely on my own control, where I get to pick and choose what I want to do?

Yeah, it may sound pretty straightforward when you leave a corporate career and just do what you want to do. I still go through an emotional roller coaster. It's been two years, to be very honest. Most of the time I'm pretty happy, content with the kind of things I have. But empty calendars, and zero inbox, is something we all dreamt of, right?

And when you have it, it kind of is scary in a funny way, because now the responsibility is pretty much on you for everything that you want to do. Back in a corporate world, you could be happy about a lot of things when things didn't go right, or didn't go wrong, didn't go right. You could just blame your manager. You could blame your team for something. Blame them for the stress.

Now I feel I'm more responsible. I just have to be more responsible, a real adult, managing my own calendar. But I think it's given me so much free time to pursue things that I genuinely cared about. One of the things I said was mentoring the next generation of entrepreneurs and students, right? I spend a disproportionate amount of time doing that at my own cost. This is not a commercial business that I run.

So I don't take any vested interest in, like, “Hey, send me a resume, I'll charge you so much.” No. Students reach out to me, I go talk to them, and if I can help them in any way, that's pretty gratifying. And that's freed up a lot of time for me to pursue things like, yeah, I would like to write a book.

I promised writing a book to my wife about five years ago, and I finally got time to do that. I have a lot of time that I spend back home in Delhi with my mom. She's alone, and it's gratifying in itself. And the fact that I can, at my decision, go spend 10 days with her, this is precious. Time is very precious.

There's one of the articles sitting on my LinkedIn post, “trading time for money.” And that's the trade-off that I actually made. Giving up on a very well-paid corporate life comes with a big trade-off as well. Sometimes I go back to it and say, did I pull the string too early? Was this trade-off too early? But at the same time, I'm like, no, time is so much more precious.

You cannot measure things in time. So it's been an interesting transition. I wrote a little bit about this in my last book as well, Without a Badge. What Next? It was not just about the transition, but talking to so many of my colleagues who were in their mid-50s, especially with the last few rounds of layoffs. All of us started almost at the same age, in our 30s, late 25s, and had great careers 30 years ago, now towards the tail end.

So many of them, everybody in good positions at Microsoft and Meta and Amazon, with the round of layoffs, many of them were out of jobs. And it was not so much of a financial challenge. We've all been extremely blessed working for these companies, how well they have taken care of us with their stock packages and everything. But at the same time, it's emotionally quite distressing to suddenly face an empty calendar.

Suddenly, as I said, I say “Amit Chopra” and then I pause because I don't have anything else to say. A principal product manager, a director here, no, nothing, right? There's no title. So the last book was actually helping these people go through that emotional roller coaster that you have to experience. But you also realize that maybe it's time to call it done from a nine-to-five perspective, and really, really thank God for this opportunity that you can pursue things of passion.

Yeah, I tell so many people that back in our 25-30s, the dream was to retire at 55, right? And now you have it. Why are you complaining, right? Of course, if you have financial challenges and you have to take care of things, obviously I'm not trivializing the fact that this could be difficult for a lot of people.

But I also know a lot of people who emotionally take it hard, because you're just used to a certain lifestyle, a certain title, a certain need to be felt in the organization. “Hey, people are reaching out to me and I'm important.” And suddenly you feel like that's all gone. So the transition is exciting. Once you go over that initial hump, I've not seen anybody go back, come back after and tell me, “Oh my God, I still think about it.”

People are very happy, but they do go through a little bit of a dip before they recover and feel so much happier, so much healthier, in terms of giving up on this very strange nine-to-five stress that also comes with these jobs. It's a trade-off, right? Everything, whether it's an engineering decision, a product triage, or anything, life is all about making trade-offs. And everybody makes them in their own way.

Your journey and your life are different, but if you can hear from me and get inspired, I can learn something from you and get inspired, and pursue those things. So that's what I do. I tell people about my stories, why I did that, and I don't tell them, “Hey, just because I did it, go leave your corporate job and feel happy about an empty calendar.” But there is a path, right?

If you're worried about certain things, kind of go through this thing and see if this is the right journey for me. Is it the right time for you now? Or if you're enjoying your life and your work, just continue, right? There's so much you can do to help people and be useful even if you have a nine-to-five job.

Yeah, one of the things that I—I've been following your content for a long time now—is that I always felt that your identity wasn't fully wrapped up in your day job. You always had additional hobbies, other passions, other interests. I think that in some ways helps, right, in being able to make the leap.

It does. Because, see, leaving nine-to-five is not a problem for most people. The problem really is they ask me, “What would I do?” And that's where I think, if you have something you deeply care about... Like, I show people, it's on Google Keep: I have a list of bucket-list things that I want to achieve. So at any point in time, if I'm in that mode of, “Oh, I'm low and I don't have anything to do,” I'm like, “Oh my god, look at your bucket list.”

There are so many things, and it immediately shifts my focus to, okay, what's the next thing I should be doing? Travel is one thing. I want to travel to so many places, I want to hike a couple of mountains, do long trails. It's a running list of random things, like, “Oh, I want to go attend the Victoria's Secret fashion show in person.”

It's on my bucket list. Watching sumo wrestlers is on my bucket list. So I look back and think, I can maybe go through one of these things and figure out what I can do next month from this list. There's enough for me, but nobody's telling me that. And so people struggle when they don't have something that will put structure to their day.

Especially in our jobs, the nine-to-five jobs, I think we have been so conditioned to follow a structure. So conditioned. Everything seems to be a ritual, whether it was, okay, we have to do daily stand-ups at a certain time, or respond to emails, responding to email at odd hours. Your mind's constantly getting a dopamine hit by doing that.

And when you don't have that, that's what causes issues. So if you have something you wanted to pursue, go back, pause for a second, and say, “Hey, I started by saying I wanted to be a chef. That was what I wanted to do.” I still have on my bucket list to go back to a culinary school and spend a few weeks doing that.

Now, it's expensive, so I have to make sure I time it in a way where my limited budget will actually permit me to do that. But it's something I have that I can look forward to. And so when I travel, I'm like, “Oh, there's something else happening here.” Okay, I'm going to Japan for something. Maybe I should go find that sumo wrestler thing that's happening.

In fact, it's quite interesting, the sumo wrestlers are actually coming to Bellevue to do a show. Though it conflicts with one of my travels, I keep on the lookout for some of these things, and that itself is quite interesting. So having something that will anchor you and take you forward, whether it is a social club, or you want to go hike, you want to pursue sports, you want to take some kind of cooking classes.

But being with people, building your own structure really, really helps with this transition. Otherwise, it can be very hard when you wake up used to a flurry of emails and pings on Teams and stuff, and nothing happens now. There was a life when, nonstop, we'd be interrupted with these kinds of things. And now you have to figure out how to manage that empty time.

Yeah, I think the key word is conditioning, and it's a big part of it. Over many decades of working in a company, if you take a step back, a lot of companies have very similar cultures, and that conditioning gets to you: the risk, the reward, the carrot and the stick, the validation you get for doing a good job, or why you do these things. That external motivation, it's addictive. And at some point you don't know how to disassociate yourself from that. It's a pretty terrible example, but did you ever see the movie Shawshank Redemption?

Hmm. It's been a while.

Okay, but one of the things that always stuck with me in that movie specifically is that Morgan Freeman's character is out of jail now, after many years in jail. And now he's working a job at a grocery store, I believe as a cashier or a bag boy. And he's been so trained after all those years that he has to ask for permission to do basic things like take a break or get a drink of water and so on.

So when he asks his now manager, who probably isn't from a background where he was in a high-security prison, the question of, “Can I take a break?”, they're like, “Why are you asking me? You can do what you want.” That moment where he realizes he has full agency over his time and that he can do all these things always stuck with me in the back of my mind. If I squint a bit, that is literally a lot of our jobs in many ways.

We're in this system for many, many years, and we're conditioned to operate in a certain way, where you have these artificial boundaries that have been placed, whether they're still real or not. You just know they're there. And so now, like a mouse going through the maze and getting to the cheese on its own, there's no maze. You can control what you want the reward to be or not to be, and the risk as well. I think many people have great difficulty thinking about their lives that way.

Absolutely. Yeah, absolutely. I think you hit it spot on with that example. That's where the conditioning part is, the validation, and having seen something in your calendar show up. And now it's like, no, I figure out when I wake up, do I take a shower? Do I go to the gym? And that's why, because you're so well conditioned, I still use my calendar to track a lot of things.

I just put them in there, so my calendar, my screen, doesn't look empty. It's very intimidating. I used to love it, but empty calendars are very scary. And so I put everything that I want to do in my calendar, and I think that really, really helps in terms of feeling that I have structure. At the same time, I have the freedom of, hey, I want to attend this meetup.

It's happening in Bellevue at four o'clock. I don't have to find an excuse to leave early. I can just start and go and attend, meet people, build new connections, look at what's happening in the industry, and come back and start my nine o'clock, ten o'clock late-night sessions. I do that a lot with India and Southeast Asia. And so it's a lot of freedom, but it's scary. Freedom is scary!

Yeah, I think in psychology and in business, they have these terms, intrinsic and extrinsic motivation. At least my observation is that if you're intrinsically motivated for the things you're doing, whether you're working for someone or working for yourself, it doesn't matter. You're going to find a way to make yourself feel fulfilled, to make sure you have purpose in the things you're doing.

But if your entire way of working is extrinsically motivated, and you have to get the guidance and the validation from others and your network around you, then I think that lack of structure can be a tailspin. I think we both see on LinkedIn examples of people at both extremes.

Some people thrive when they're given unstructured time where they can take all their passion and expertise and channel it into something constructive that makes them personally happy. And we've seen many examples as well of individuals who, when they find themselves with an empty calendar and unlimited options in what they can do, free-spin.

Yeah, absolutely. Hit it on the nail. Hit the nail on the head.

All of it works. In the world of AI, whatever the output is going to be, there'll be some solution to make it happen the way you want it to be. So speaking of AI, one of the things that you and I talked about before we had this call, like a casual chat, is that I always felt that we got into tech at the right time.

Any time too early, it would have been way too early to take advantage of all the great growth that's happened in the industry. And as we're seeing right now, because of AI, there's a heavy amount of pressure for how can we get more done with far fewer people? How do you feel this is going to transition how you're approaching your guidance and mentorship for students and for people early in their careers?

Yeah, so this is one of the things I've spent a lot of time thinking through, talking to a lot of people, other experts who have also been thinking about this. One thing you said, we were in tech at the right time. I think people who are coming into tech now are also coming in at the right time.

When they look back on this ten years from now, they'll be like, “Wow, we were in the best position because we had a tool called AI at our disposal that made us so much more productive in the things we wanted to do.” Now, the challenge really comes down to how the education systems have conventionally been created. Look back at how almost everything, or nothing, has changed in the last 50-100 years.

Fine, MOOCs came out, and suddenly you could learn online and do something without going to a school. That itself was a shift that was painful when it started coming: who would do that, and why would you do that? You should go to a classroom. No, people are learning a lot of stuff today also. When I spend time going and talking to students and academia, they're still teaching things that were taught 30 years ago, with the same rigor.

And the relevance of that is decreasing. I don't think the education system has been able to catch up quickly enough. It's just how they are. I spoke to some people, and it takes 18 months to go change and get the approval processes done to introduce a new course or something. And so all we do is spend $30 a student and give them Coursera subscriptions: learn something new. So education institutes haven't caught up.

At the same time, I think technology is also moving much more rapidly than in the last 20-30 years. Again, it's that classic hockey stick curve: things were moving up, and now it's too fast. And so there's a dilemma on both sides. Academics are not able to catch up, technology is moving too fast, and the gap is increasing much faster. Any of our traditional approaches which said, “Fill the gap,” are not going to work.

This is something I've had conversations about with a couple of universities. Three years, four years ago, not that far away, you could come out, you can teach them whatever you are teaching them in the university, throw in a React course for them, and they are ready for a professional job. Or pick any. I just picked on React because a lot of people got jobs as React developers very quickly.

That part is changing, that mindset is changing, and universities and students are still asking me the same question: how do I fill the gap? I mean, you can't fill the gap, because the goalpost is moving so fast that you can't even define what the gap is. So now it comes down to building an understanding of what you can do with AI that makes you effective enough. AI is not your enemy.

I mean, you can see it as that. This is how a lot of stories keep coming up. “Oh, AI is taking your jobs. AI is taking your jobs.” And it stays in students' minds and in a lot of people's minds. Obviously, there are many factors behind it. Jobs have not grown as much. AI definitely is one of them, but not the only one. At the same time, AI is also a tool.

It's an amazing resource that you can now use to build your skills, to learn things, to solve problems. And I have been thinking in terms of not telling people, “Oh, everybody should become an entrepreneur.” That's one end of the spectrum. It's like, “Oh, if you can't find a job for somebody, you start your own stuff.” Hey, that's an easy answer. Very easy. But entrepreneurship is not for everybody.

There are a lot more attributes that are required: the ability to take a risk, financial risk, rewards, recognition. But at the same time, I tell people, start building a builder mindset. Become a problem solver. Then AI becomes a tool to an end. And there are so many problems that still exist in the world that you will not go hungry, you will not have the need to go find a traditional job.

If you can spot a problem, you can use AI and solve it. So the transition is not straightforward, unfortunately. There is no quick fix, no “here's something you fix in the gap.” But if you change your mindset from “I wanted to be a developer and I'm programming” to thinking about the next step beyond that... You were being a developer to solve some problems. Coding was not the end. It was a means to an end.

So start thinking along those lines. That's where my mind has been. I talk to a lot of people to see, am I missing something? Am I maybe too biased, because I've always had this product mindset? One of the talks I've done at almost 30-40 universities and events in the last two years is about shifting focus from “I'm engineering for the sake of engineering” to “I'm engineering for the sake of actually solving a problem.”

You kind of shift right on that spectrum of what more I can do with AI. And if students build those skills, if academia can teach them how to be thinkers, I think we will solve a lot of these issues. But the transition is not going to be that straightforward. People still want conventional jobs, people still want traditional degrees, and that's going to stay for a substantial amount of time. So that's where my thinking has been.

Would you say there's value in someone getting a computer science education in college today?

Yes, to a certain extent, today, yes. There's still a lot of need, personally, for people who understand how tech works under the hood. I think that is going to reduce significantly over the years. A lot of the stuff I had to learn—there was a time when I had to learn assembly, which was the bare metal, the bare bones of what these machines are running. Everything gets down.

We didn't have to wait for compilers and stuff. We just went to the metal and wrote exactly. Those things went away. C came up and abstracted quite a few things. C++ abstracted a lot of the memory management, the pointer stuff. Then Java came. Then newer languages are coming. Visual Basic was like the best abstraction ever. We could save a button and write button.click, and there are millions of lines of code running behind that.

Did any of us care? Some people did. Some people in the Visual Basic team absolutely did, because they had to make sure it would respond in some milliseconds. Do you have to care about that? No. So it's the same thing. I think we still need a lot of people who understand the nuts and bolts, but I think it will gradually start to decline. I think the whole evolution of programming languages was for human beings.

If you think of why we abstracted: I programmed in COBOL. I used to work at Digital Equipment Corporation Limited, DEC. That company is gone. They had machines the size of refrigerators, less powerful than the phone that we have, and we wrote programs in COBOL because we could all understand that. I don't think VMS cared. At the end of the day, it was all running in assembly. So we'll see.

Maybe that shift happens back to a language that these machines create, highly optimized, where they know exactly how to communicate. A lot of things were being done to make our life as programmers easier. You can see even today, I feel there's a very conventional way. We started with an IDE, a debugging thing, a deployment thing, and these things are what we are optimizing for.

Everything we're building, an agent, is like, "Oh, we'll replace the human and put something there." Then interesting things happen. It's a complete shift in how people think. Classic example: Windows Phone. When we were starting to think about the UI, all we thought was we have to put in a start button that works beautifully like the big Windows. I remember my daughter was maybe a year and a half, two years old.

I have tons of Windows devices, and she would have a stylus and tap around the screen, and once a window goes away, she doesn't know what's going on. But she would keep tapping around. Then I actually got an iPod, one of the earlier versions that started having apps on it. It was the iPod which had the apps on it, and she had a squishy ant app which she loved.

I could literally see that once she had that phone, which does not have a start button, which worked with pinch and zoom, and she could swipe things, she never touched the Windows Phone again. It was so groundbreaking of us optimizing something for a behavior when Apple came and completely changed how we used phones. So that's where I feel a lot of how programming is seen is how it's evolved over the last 55, 60 years.

I think it will probably take some kind of shift as more and more of these agents just figure things out. Everything's happening under the hood. There'll still be a lot of specialists who will be building those very intricate systems. But for most people, I don't think what they are teaching in computer science is going to be as relevant from a career perspective.

I don't know if the timeline is six months, if the timeline is five years, but I can see that change will happen, and the sooner people prepare for that. If you're going to CS today, that's totally fine. I don't think you'll struggle finding a job, because you're foundationally really solid. You're still one of those people who understands how things work on the nuts-and-bolts level.

And I can also tell you that the spectrum of space is so large. I was mentoring a company very recently, and I saw the apps that they're making. I almost wanted to say, "You're taking me back to the 90s." They had a web page, and you would upload pictures, and there'd be a dashboard. But there's a market for that. That's not going away. There's nothing wrong in that.

There are real problems, there are real systems in place for which we need people with these skills. So yes, you may not get the most glamorous things, and if your expectation is, "I'll come out of college and start with a seven-figure salary," that's going to be smaller and smaller. But if you're doing CS, you will have a job. You're learning some very strong foundational skills.

But just don't depend on the fact that you came out with a CS degree and a recruiter will be waiting at the door. You will have to make sure you're ready for what the market needs. That's the main thing. Your understanding is going to be important. That's why I try and help people. I tell academia to help people as well. But it's not the doom-and-gloom thing. Don't go to the extreme.

Science fiction has polluted us enough to always feel scared of these devices. And on the other side, it's fascinating to see how some of these things are coming out. But there's a massive market of traditional software, traditional companies, traditional jobs that are still there, and that's not going away anytime soon. So they should feel absolutely comfortable.

Yeah, because I think the classic tech cycle has always been, you mentioned problem solving is the big part of it. You needed a certain number of people who are great at identifying the problem and defining what a solution to the problem could look like. And then you had a large group in the middle whose job it was to figure out how to make that solution being proposed a reality.

And then you had a small group at the end to make sure that we tied the loop, that the problem is actually being solved. It's done with high quality, high taste, and all the right things to make sure we meet the business need and the customer need. I think with AI, what we're seeing, though, is that the big middle part, on how do we take a problem and make the solution, is now being entirely automated and done by AI.

And to your point, we've spent decades optimizing everything for the mouse, keyboard, and screen, and programming languages, because that's the best way a human could interface with a computing device. When all of that goes away, AI can create assembly, can just create zeros and ones directly. It doesn't matter anymore, because that middle layer is entirely automated. All you care about is: is the problem worth solving, and is what is being generated the right solution to that problem?

And the question then becomes, how big of a market is that? Are there enough of the people who were used to being in the middle, who defined their identities and their skill set in being able to take a problem, break it down, and actually solve it themselves, to figure out which of the two buckets they can fall into? They're identifying the problem, or validating the problem and making sure it works.

I think that's the tension you're seeing in many ways, because of the speed at which it's happened. I think it's quite fast. ChatGPT was what, maybe three, four years ago? Between then and now, the massive amount of change in how we even think about solving problems is probably the fastest I've ever seen in an adoption of a technology that can disrupt the processes we know.

Yeah, and not just software. I think it's disrupting so many other places at the same time. And I think that's something that scares a lot of people. As I said, I go through this motion of, "I love AI in the morning and I hate AI in the afternoon." It is because of that. There's a part of me that thinks, saying that I'm at the tail end of my career, I can abstract and think a lot of things.

But if I put myself in the shoes of somebody who's literally a 20-year-old getting into college and seeing all this uncertainty, which end of the spectrum I am. I can be a thinker because I have had 30 years of thinking experience, because I've gone through that. How does somebody who's 18, 19, 20 years old claim to be a thinker? I'm not there. When we hire early-stage product managers, I'm looking at their execution skills.

I'm not asking them to define problems and define solutions. That's the next level. You have executed well; now here is a spectrum of problems, pick a solution. You start with picking the best solution. Can you trade off different solutions? And you come back and say, here are ten problems, can you trade off which one to pick? Then you become more senior and you're like, "I can identify these problems at a certain scale for my organization."

And then you get even a step higher. I think that step-level change is concerning for many of us who've gone through five or six of these. It's easier to say, "Put on a builder hat and start thinking," but there's not much training or experience. And that's where the genuine issue is. That's where I feel I don't have a real black-and-white answer that I can offer them even today.

I get nervous, but I also tell them at the same time, don't get all the way into the doom-and-gloom scenario. There's a massive market out there. You want to perhaps just go and work at big tech, and that's absolutely fair aspirationally, but hundreds of companies are building software. Nothing on the ground has changed, and there's no real decline in any of the job applications that are coming out.

A lot of statistics are being published. So let's go look at the raw data and see what's concerning here. I've actually been working on something recently where I have gone, and again, I've used a lot of AI to help me out here, to go back and dig through how Microsoft, Google, and Meta have changed their hiring stance for entry-level people. I went through job descriptions posted in 2006, 2008, very early days.

Microsoft never even had a product manager category at that point. We were all program managers. But Facebook and Google came up with this new investing title, and I looked at how they've changed the hiring. I've been literally looking at the fine print of when the word AI actually ever got introduced. And it's amazing, because now AI can do a lot of analysis for me.

I've got tons of these job postings that I've picked up from the Internet Archive, and I'm trying to see what has changed. Interestingly, not a great amount of change. The foundational skills that they're still looking for in people haven't changed. Of course, they're looking for AI proficiency, because if you can't use a tool that uses AI, or you don't understand it, you're not ready for the market.

That's why I said you have to be prepared. I tell people to understand how AI is used. If you're a developer and you haven't used Cursor or a tool that uses AI quite deeply integrated, you're going to be outrun by somebody who knows that. It's not like AI is taking your job. Literally, the cliché statement: somebody who knows AI is taking your job. They'll be more productive.

When I hire somebody who understands AI, who understands how to use these things, I would naturally gravitate towards that person. And I tell them a lot, like in college, identify problems. There are so many problems in the world. Pick up something that bothers you on a daily basis and just use a machine, use an agent, to code something, prototype something.

It doesn't have to be something that gets you a YC application, but it's something that you will actually feel really proud to talk about. Having done something demonstrates your ability to take on challenges and be a thinker. Plus, you'll also realize the number of trade-offs that you'll make, even if it's a two-month-long thing that you're working on. That's the kind of skill that you need to develop, that conventional academia doesn't teach you.

Conventional academics is, we grade you against a rubric. You do this exam, you get a grade, you get results. Real life is not like that. The moment that finishes, you're competing with people, you are being evaluated against five other people, you are stack ranked in every job interview you go to. I tell people, you could have been perfect in what you did, but there was somebody who was better than perfect.

They demonstrated something that made them stand apart. It's not about, "I get 90 and the other one has 89, so by default I'm number one." Those are things academia has to focus on, we have to focus on, so that students who are coming into this career feel that what they're learning is still relevant, but it's not. It's table stakes. It's not going to get you a six-, seven-figure salary to start with.

Yeah, I think it's all about how you approach the problem and how you think about it. Your example earlier, we talked about how your daughter used to use a stylus for the Windows Phone, but then once the natural touch on the iPod became a thing, that's going to be the future. And from here on out, it's never going to be natural to ever think about a stylus and all of that.

I wrote about this pretty recently, where I said one of the biggest things that limits me personally in my use of AI is that I always have this artificial conditioning limitation of, "Oh, but I will need resources to be able to do this," or "This is difficult because it requires X amount of things to invent and all these capabilities." I always think in terms of what's the goal and what are the resources available to be able to do it.

But with AI, that part completely goes away. It's almost an example of whenever I look at technology, I always go back to the old dial-up days, where you actually had to launch the, I even forget what it's called, but it used to be that little tool in Windows 95 that had the connect button and the phone number you enter for your ISP. You hear the sound of the modem connecting and all of that.

Yeah, yeah, yeah. MSN dial-up. The MSN dial-up was there.

And then you go there. That's right! All of that. Whenever you look at a problem, for a brief millisecond, I start with those first principles of my earliest experiences with computers and then go all the way to right now. So I feel like people entering the industry today do not have that same limitation. They only see the unlimited possibilities of what you can do with AI.

So their default answer is, "Of course, you can do this, and you can do that, and you can do other things," without placing artificial limits on what can be done, which is something that I struggle with. Again, going back to conditioning, after so many years of knowing that you can't get your P1, P2, and P3 items done. You can only get a subset of your P1s done, to be very close in prioritizing it.

But with AI, you can get all of that done and then some. So you have to train yourself to think about the end result beyond just the MVP, which might be a watered-down version of things.

Yeah, and I think some of this is also interesting, maybe a little more controversial as a thought: a lot of the jobs and a lot of things that we end up doing were because of inefficiencies in the system. We used to have a note taker in a meeting. Nobody wanted to be that person who would sit down and record these conversations. I attended dozens of ship rooms, and my life existed because I was a great note taker and I could chase people and remind them.

That job's gone, right? And why would I want to do that at this point? But if that's all you developed your skills and were good at, obviously, be ready for that change, because AI will do it so much more efficiently. All meetings now get recorded. Transcripts are sent around, notes are sent, Asana is updated, tasks are created, managers are reported, with so much more efficiency.

But you can complain and say, oh, I had this great job as a ship room czar—that's what we used to call ourselves—to just remind and act people. But that was a skill, right? The fact that I could remember 15 action items from a meeting when nobody else could gave me a very interesting position in that particular room. But was it the best use of my time? Was it the best use of the company's time?

And many of these jobs, like, we hired program managers. There used to be a joke that you hire three and you have to hire a fourth one just to manage the three program managers. It's interaction between them, right? PM roles came to solve a very specific problem, and then we created yet something else. And there were so many people that got hired because at that point it was required for me to run my product engineering successfully.

We have to make sure communication is happening smoothly, things are not slipping through the cracks. And people got hired to do those jobs, right? Things changed. Technology came in, and it's like the industrial automation. This kind of started taking a lot of these mundane jobs away. Now, I could, as I said, feel too passionate and say, oh, I've just lost my job because AI took it away. But look at the grand scheme of things, right? It was probably not even necessary to do that.

And that's happening in a much bigger spectrum now. I think that's where it gets a little more concerning. Lawyers come and talk to me like, oh, we used to do these, write up these [unclear], and that's what I did for the first two years of my job. And AI does that in a fraction of a second now. Where do I build my skills? And I think that's a genuine concern.

Medical analysis is the same thing. I upload a lot of my reports to ChatGPT, and it does an amazing job with analyzing things. So these things, like the resume thing, I know when I was in UCLA doing MBA, companies had approached us. They said, we'll charge you $1,200 to format your resume and make it ready for a job. And now pretty much anybody who has access to ChatGPT can do that reasonably effectively.

I would say reasonably effectively because, yes, those people were skilled and they did certain things in the right way. And it is interesting how it's seeping into so many parts of our life. Now, whether we like it or not, that is the future. So let's just make sure we are ready to adapt to it in the best way to our advantage. Sulking and complaining is not going to help.

No, exactly. There's a great cartoon or a comic I read a long time ago, which is really about—so AI generates all this content, and then the person who's receiving it uses AI to summarize it back down to your essential thing. And in the middle, we just had stuff that was created just from a historical reason point of view, but it doesn't really move the needle anymore.

Which is also true in the academia space. This is what's happening. Professors are using AI to create quizzes and things, and students are using AI to solve those quizzes, and professors are using it to grade those quizzes. There's actually not much of learning happening there. And that's what I said, right? We're trying to optimize for the things that we are used to, and that's not how change happens. And in this whole process, the students actually never learn how to use AI effectively.

I did a product management workshop a while back for a university. I'm a big proponent of how Amazon does the whole working backward thing, right? I think the two years that I spent at Amazon learning how to build products were, by far, the highlight of my career in terms of being a better product person. Every approach, everything they had, started by writing a press release: articulate what you want to build versus making a flashy PowerPoint or coming up with a swanky, working prototype.

None of them are acceptable. No, you put your thoughts down on a piece of paper and write, basically, a press release. Ideate your product. And it was very hard because now you had to really think and put that on paper. It's an extremely difficult thing. So, obviously, things have changed there. They also use some automated tools to create these artifacts of PRDs and come up with user stories and press releases. Totally fine.

So we did this as part of the workshop, and I said, let's write a prompt and come up with a PRD. And we just went through this quick exercise of identifying a problem in the class. And what I ended up then showing them was—I basically said, people, I don't want to actually see your output, whatever you came up with. That was not the whole point of the exercise, even though I made you do that the hard way.

All I actually want to see is, let's discuss your prompts. And we then went through this exercise of looking through prompts, and I had these four variations of prompt that I had created. One was like, create a PRD for this particular problem. And ChatGPT does a pretty decent job, right? Even with a two-line prompt, it would do its best to impress you. It comes up with almost everything it thinks is right, or hallucinates in the process.

And then I started showing them more advanced prompts on how you put constraints, right? How do you put artificial limitations? How do you nudge it to create content in a certain format? And in these four prompts, you could literally see a difference in the output.

So that's what I felt: the AI literacy is so important. It's not just that you have access to a tool. You can give it a really basic-looking prompt, and ChatGPT and Claude and all will completely impress you with an answer sounding so realistic, will have literally no merit. So how do you, A, prompt the system, then how do you go and validate some of the things that it has actually produced for you, to make sure that it's ready?

But people could see that what we did over weeks of time could now be done so quickly. Just analyzing customer feedback. We used to have UserVoice back in Visual Studio. We would literally sit down for hours, going through the feedback and putting them together, putting them in a spreadsheet, trying to stack rank them and say, oh, these are the two or three problems we should put for the next version of.

Those things can be done literally on the fly, in real time, in a fraction of a second. Say, hey, tell me what the customer is doing, get some stack rank information. So, things make you so much more productive from that perspective as well. But using AI, you have to also teach people how to use it in the right way. And I think there's very little knowledge that is out there because, again, things have moved so fast.

I may think I had the four prompts that were really good to teach people. Maybe there's a fifth or a sixth thing. Maybe Claude and ChatGPT already have added a few more features, because they are also improving so fast. I mean, the fact that even three years ago when ChatGPT came out, everybody just complained about hallucinations and it's not factually correct and it doesn't know this, it's cut off on this particular date. And so we can't trust a lot of what it's saying.

To the point that today, there's so much you can do and feel quite confident that they have done the right things, they can reason. When the Chinese model came out two years ago, I was in India that day, and suddenly, the only thing that triggered my attention was the stocks completely tanked that day. The tech stocks took a huge beating, and I'm like, what is going on here? And it turns out they said, oh, this Chinese company has come up with this new model, it can actually reason.

And like, what does that actually mean? So I'm like, okay, let me log in, try it out. There was something I was using ChatGPT and Claude for. There was a stock trade that I had done. I was learning trading at that point, options trading. And one of my trades had gone south. It wasn't as I had expected. So I'd taken some screenshots and pasted it in ChatGPT and put some comments and said, figure this thing out, this trade is not going well.

It was giving me some answers, but I could make out this is not correct. What it's suggesting does not seem right. Then I said, let's see what this reasoning model can do. And I pasted the same prompt I had in ChatGPT into this, and then it starts actually showing its thinking. This was the first time a model had actually started typing how it's analyzing your thing.

And then one of the things it said, the person actually says they are selling a put, but it looks like they are buying this particular option and not selling it. Let me clarify this thing. And it actually comes back and asks me, are you sure that what you said in your prompt is correct, but it's not actually resonating with the screenshot that you attached? And that was a jaw-dropping moment for me in terms of, oh my god, these things are now showing signs of certain intelligence.

Like, this is how intelligence works. And it actually found a discrepancy in what I typed versus what I had put in the screenshot. And none of the other tools were able to pick that up. They were just answering me exactly based on what I had told them. And obviously within three or four weeks, almost everybody threw in the whole reasoning thing, but that was a huge shift in terms of what AI could do now. Gemini added the research feature. Now it's pretty much everywhere.

Default.

You go to Gemini, it's there. You go to any tool that you have, record your meeting. Nobody's taking notes these days. Everybody has an AI actor that joins in and then sends all the follow-up notes. And it frees up. Even for me, I felt, even during interviews and all, you're in constant pressure, right? You're scribbling your notes, and half the time you're listening, and half the time your subconscious brain is filling things up without us even realizing that. So AI kind of completely changes the game there.

So the spectrum is so huge that we need to make sure people can understand how best to use it to be more and more productive. And I think that's where I've been spending a lot of time with academia: okay, let's figure out what we should teach, how we should teach curriculum, what else can we add without disrupting what you already have to teach.

And you still have to teach them the foundational stuff, but can we get them ready faster for what they would be expected to do in the real world quickly?

Yeah, I think with Boris Cherny, one of the Anthropic Claude Code creators, he mentioned the best way to work with AI now is to treat it like a human colleague. Given the level of context you would give another person, and then they're going to be very effective. And I think your example, referring to the DeepSeek model that came out like two years ago, is very similar to that. When you're talking to a person, how much context would they need?

And as the models get better, the person you're talking to, you can think of them as—it used to be that they were early-in-career engineers like three years ago. Now they're very effective at their job, domain experts in their field. Who knows six months from now how much more advanced that individual can be? But if you prompt them in a way that gives the most out of their output, you're going to be more effective at solving the problem you set out to start in the beginning.

Yeah, I think for me, I have these very aha moments sometimes when I watch some of the videos. The first one was—I don't know if it was a video that Google wanted to put out or it got leaked—a YC demo or something that they were doing. A small clip came out on the internet, and they had a robotic arm, and this guy went, and there was a table of toys sitting there. This was like five years ago or something, right?

There's a table of toys and there's this robotic arm, and this person says to it, “Go pick up the extinct animal.” That's pretty much the thing it was told. And this thing took a picture or whatever, went and picked up a dinosaur, and pulled it out. For me, that was a very aha moment. It's like, my god, this thing actually had to figure out, these are the animals, this is a dinosaur, it's actually extinct.

And this was for me—that video may be sitting there, but I don't know, it was hush-hush for a while or something. But it was one of those moments when you feel that the machine suddenly got smarter than just following a very set sequence of commands. It is figuring something out without having to know. That was one.

The second one was when Devin came out. This was pre-Claude Code, pre any of these fancy tools that are becoming so common. But when Devin came out, they were actually asking for $500 a month or something. I don't know how many people ended up using that, but conceptually, I could see that this is going to be a game changer.

Devin is now part of your Slack. Devin has a machine available itself. Devin will read all your docs and you can interact with that. That is when I felt that there's going to be a massive shift. This was also three or four years ago when something of this nature happened.

Then we also had a lot of these add-ons that came on ChatGPT. I'm completely forgetting the names, which actually used a lot of RAG. Basically you'd say, give me a goal and I will solve it for you. So it would recursively keep running and putting things in the memory. There are three or four. Baby AGI, yeah, that was the one. Baby AGI was an open source tool which kind of started to demonstrate very early signs that reasoning can happen. It can question itself. It can actually go solve for a goal.

And the most recent one, which is fascinating, which I saw only last night, had the same aha moment. But did you see Elon Musk's flying car demo?

Of course, yes, I did.

So, again, I would not—those are the kind of things that bring the kid-in-the-candy-store thing for me. It's like, okay, this is something that's going to change how we look at a problem. This is where I felt I think he's done better than our Google moonshot thinking of, hey, I'm going to flip a car upside down. The first time, the car went halfway through—I'm scared of heights, my heart was already sinking—and it came back.

Then it went up and did a full flip and it landed. Okay, that's not how we'll see cars in the future on the road, flipping, but I think it just opens up doors for something new for a lot of creative thinkers to not think on top of that. What an interesting model, right? It's only so much of liquid; you don't have to fill anything specific. You just use eight SpaceX jet engines and come up with something creative.

But it's one of those starts of an idea that I think a lot of people will be like, oh, maybe we can solve a different problem like this. Maybe it's the things we see people in Dubai do, where they wear this backpack and they're moving around in water using that. Maybe we can do it in a safe space.

Or I was telling my wife the other day that I spent a lot of time reading Hindu mythology, not from the spiritual perspective but from the hidden science that I always think it has. And there's something called the Udan Khatola, the flying saucer. It's been talked about so much in our mythology with so much accuracy and so much detail. And this, for a moment, actually just said we are on the way of creating that.

We are finally—the future is taking us 10,000 years in the past, when we had a lot of these things which you could never describe, whose science basically vanished from this earth. But these things are going back in the same direction. We will have seamless transportation options, completely free of all the fossil fuel and things, and we'll come up with interesting ways of modifying gravity or doing things in different ways.

Like flying itself has only existed for what, 120, 130 years? Maybe even less. And we take it for granted. 200 years ago, nobody could envision flying from one place to another. Just the fact that you had a smarter cart or a train was a big achievement. So that's where I think these pivotal moments do happen, and you can only appreciate them 20, 30 years down the line when you go back like, oh, I can see that, right? This is a massive pivotal change in the making.

So these four things, three things, have been quite fascinating for me, and I always spend at least 20-30 minutes on YouTube trying to find this kind of stuff. It's on my calendar, actually, to go watch a certain set of channels on YouTube every day and see what's the new thing that people are talking about. What's the exciting thing that's happening?

Yeah, today's frontier research and novel thinking is like tomorrow's commoditized solution for your everyday problems. Just like how NASA and the space missions ended up creating mundane things we take for granted, like toothbrushes and toothpaste and things like that, because the technology trickled down and applicability was found, or product-market fit, in more commoditized areas.

People who are even being born today are being born in an amazing time, I think better than what we had. I struggled getting off in this field. My wife jokes about this. When we met each other, we used to take a bus back from office. It was one of the chartered buses, and we had a long commute back.

And I would sit down with a really fat book. It was the Windows 95 Charles Petzold book, and I was preparing for my Windows 95 certification. That's all I had, a big, fat book. I'm literally spending all the time reading, and that one book is all I had. It was so precious to me. And now, if you look back, or look ahead, people have so much resources. If I have to learn something new, it's completely different how I would take that approach.

So I think things have become easier for people now. A lot easier. So if you're coming into tech at this point, I think you're still in the best time. And people who come in 10 years from now, it's going to be a better time. I'm one of those people who feel that the future is better and brighter than what we had in the past, right? So any next generation should benefit from things we have done.

The generation before us did a lot of interesting things. Dennis Ritchie and all created the C programming language for us, right? So we didn't have to struggle with assembly. People in our Microsoft created .NET and VB and VB.NET and made life even simpler for so many other people. And so each generation leaves something which they should be proud of and a good stepping stone for the next generation.

And I think today, people, if they have to be and can be, will be more creative and more productive by orders of magnitude. I should actually feel envious and jealous that I was born like 20, 30 years earlier than I should have. That's how I would put it.

Well, Amit, on that note, I think it's a great stopping point for our conversation. Thank you, that's fascinating. And we'll have a round two at some point where we figure out what are the new things that happened now and then.

Absolutely! Let me know. My calendar is empty, all you have to do is just let me know.

I'm still in the nine-to-five conditioned race, so mine is not, but one day, one day, you'll be an inspiration for me and maybe others as well to make that leap.

Yeah, absolutely. Perfect. Thank you, Kirupa. I appreciate it. Yeah, take care.