Beyond Code Assist: Driving Impact as a next-gen Developer

00:15 Alright. Alright. Alright. Alright. If you've ever seen me before, you know what's coming next. You make me do air squats, Mark. I swear to God. I need everybody to stand I'm kidding. No. There's no air squats. Here's what I need for you to do for me. I want you to rub your hands together.

00:33 And I want it to sound like it's sandpaper all through the room because our hands are rubbing together. And then on the count of three, we're gonna clap in unison. One, two, three. Oh. Yeah. That's what I like to call a thunder clap. I like to bring the energy. Thank you all for being here.

00:56 Thank you for just giving time and space. Today has been outstanding. Can we give a round of applause for all of the speakers so far? Outstanding. It's unbelievable to be at an event with so many amazing speakers filled with knowledge, filled with everything. My time is going. You can pause that because I'm gonna go until I'm done.

01:19 Y'all go ahead and pause. They got this little iPad up here. I'm not looking at that. I'm not looking at that.

01:27 These two letters have changed the way that we think about the world. We had a plan. We all went to either a coding boot camp. Maybe we were self taught. Maybe we spent time in university. And we were like, we are going to be engineers for the rest of our lives. And what are we gonna do?

01:48 We're gonna make apps. We're gonna build websites. We're gonna do stuff. And then this led these two letters just kept popping up. And now it's to the point where you can't even look on the Internet without seeing these two letters now. You can't go anywhere.

02:06 I was once at a park with my at the time, let's see, he was about four years old, my four year old son. And, you know, all the parents congregate on the side looking at their keys like, oh, don't hold yourself, John.' You know what I mean? It's really intense.

02:17 And one mom goes, 'Hey, what do you think about chat GPT? And I'm like, you don't even know me. Like, I've never even seen you before. And that was the first question out of her mouth because this term has become so pervasive. And but for us as engineers, as makers, as builders, something else has happened.

02:36 It went from just AI to AI fatigue. Now, this is a safe space. You can tell me how you really feel. Anybody tired of hearing about it? Be honest. But let me tell you why you're tired of hearing about it. You're not tired of the idea that we are building new technologies that can change the world.

02:54 You're not tired of that, I bet. But you're tired of this. Check out our new seven gazillion programmer and model that outperforms PhD rocket sciences at every level. You're tired of hearing that because literally, you cannot go a week without a new model drop. It's some of the stuff people don't even know what it means. 7,000,000,000 parameters.

03:16 What's that? Like a long function? That's what we're talking about? That's that's a cold smell. That is a cold smell. What are you talking about? But that's not all. I bet you're tired of this. I bet you got a lot of fatigue of this one.

03:33 You're tired of hearing it because you fought for these skills.

03:39 You worked hard. I know some of you maybe went through a career change to get into this field. I met a teacher today, a former educator who's now into tech. You know how hard it is to switch as an adult? You fought hard for these skills. And now people from the top levels are like, this is it.

03:58 Coding is cooked. We got a new chip that's going to run so many models. We won't even see web developers in three years.

04:07 I'm tired of hearing that. I'm tired of it. I can't let me look. Somebody is going to see this and I'm going to get an email from work tomorrow. And they were like, what are you talking about, bro? You know where you work. Some of y'all are like, I know where you work.

04:19 I know, I know, I know, I know, I know. But here's the truth of the matter. Things will change. If you're taking the position that all of this is a hoax, that's incorrect. I'm sorry to tell you. If you're taking the position that this is a fad, that is incorrect. And I'm sorry to tell you. Okay, friends.

04:43 That's not the position to take. I want you to think more like this.

04:49 We can control how they change. We have influence. We are the builders. We are the creators. The models are trained on our work.

05:02 The stuff that you did. Models don't know how to code on their own. So you have influence. The story does not end here.

05:18 So let's talk about these two letters again. But now there's another term. Now I'm not so tired of this, but nobody knows what it means. It just sounds provocative.

05:29 Get the people going. You see me? It's the idea of an AI engineer. And so let's talk about it. Now first and foremost, as I talk about what an AI engineer is, I am not to be held responsible because these terms probably will change. But at the time, friends at home, thanks for hanging out. Listen.

05:45 At the time of this recording, alright, this is what I think it means. But do not DM me that I was wrong. Alright. This is what I used to think an AI engineer was. Somebody who would build and develop with ML algorithms. Right? That's when you think about an AI engineer.

06:00 You're like, man, machine learning, you know, all these algorithms, all this math. They were, like, mythical creatures that, like, you knew they existed, but nobody ever met anyone. Right? Because they were all because where were they? They were cooped up in research labs. Right? And they were just doing new stuff.

06:15 And it's like, yeah, I know y'all out there. Right? Like, I got a recommendation engine now on Netflix. Like, somebody made that. Well, it wasn't me, so whatever. Then I feel like we're migrating to a new one. Okay?

06:30 Uses code assistance, which I have no problem with that. I want to talk about that. Uses code assistance

06:37 and Vibe codes everything into existence. When I first started working on this talk, that second part didn't even exist. That's how much this world is changing. There was no such thing as live coding when I started thinking about these ideas. Because here's what I thought about the most.

06:52 What do we need as a community, as a developer community? So I'm putting stuff together, and then I hear about vibe coding where you don't even care what happens. You just say what you want and throw caution to the wind. Just, Alright, whatever happens, it breaks. Put it back in the L. L. M. The L. L. M.

07:07 Will fix it. And eventually, you create something complete. It's amazing. I don't really understand it. It's not the vibe I'm looking for. But that's okay. But then there's this third one, which is something we're going to talk about. I think I have an answer for the third one.

07:23 A more founded answer that's gonna give us some more direction.

07:29 But as you look at AI systems and you think about these things, you probably say stuff to yourself like, sure, it can code,

07:40 but I'm faster and smarter. Anybody ever said that? It's okay to be honest. It's a safe space. I see some of y'all was in here lying. At home, you can't see their hands, but some people are in here lying because they know they said it. They know they did because I said it.

07:52 Because I looked at those early prototypes from OpenAI when they were showing those Python functions. Right? Because that's how it all started in the public space. They started showing Python functions. I'm like, that function is bad. Look at the parameter names. It don't make any sense. And then it got better. Then all the models got better.

08:10 Then it became well, it might be faster than me, but I'm still smarter. Because I wouldn't make the same mistakes. Knowing darn well I'm lying. Because, listen, the person here who has never made a coding mistake, raise your hand. Exactly. Now your hand stayed out. Kent, don't you play with me. Kent, don't even put that hand up.

08:31 But then, guess what happened? Time went on. More parameters get added. Umhmm.

08:42 And now it's, I'm not smarter than it anymore. Right? I was smarter than it, but now I'm not.

08:52 But this is the thing that I'm hoping for, that I'm needed. I'm going to give you a little bit of a plot twist. You are. I'm going to tell you why. But you are. So let me ask you a question. Who's afraid of the big bad model?

09:10 Me? Mark?

09:17 I wasn't gassy to raise your head. I'm just telling you the truth, Mark.

09:24 At first, which is what drove me to this talk, Which is what made me reach out to Kent. Like, bro, I have something that I really think that people could benefit from. Because I was tired of being scared. It's exhausting being scared. It's exhausting being worried.

09:42 Is no one else exhausted about being worried about your job and about your future? Am I the only one who's exhausted? I know I can't be. I'm just outside. So I started thinking, so what? And and why even be afraid? It's because they could just do things. And when I say that,

10:05 when I say that, it's because they kinda can. I can tell you examples when I've been man, they really need to turn this thing off because I'm not looking at it. Alright. Cause I've looked at examples.

10:16 And I look at the, you know, and I'm like, hey, you know, give me, you know, 25,000 rows of data in this shape and format. Don't don't make any rows the same. Make everything unique and it can just do it. Then I'm like, hey, my wife's not going to Las Vegas. Give me some itinerary.

10:31 It can just do it. So it can just do things. But you know what I figured out? It's a big secret, I'm gonna tell you. Y'all ready? Do your hand like this so you can hear me. Let me see. Y'all ready?

10:55 You can just do things. And you can. You can. Because you are a problem solver. Since the beginning of your time in this industry, you have had a pattern. You see a problem, and then you solve it. And you think about it. You digest the parameters and you figure out what has to happen next.

11:16 That is what you all have been doing since you started in this game. Why do you think that that's suddenly invaluable? I'm sorry, why do you think it suddenly lacks value?

11:28 Because we've been here before. When I first started my career, you know when we wrote code in? C plus plus And then all the Java developers came around, and the c plus plus people were so mad. What do you mean you don't have to manage your own memory?

11:46 Java developers are like, I got garbage collection.

11:51 AI vibe engineers, like, I got vibes. Okay? So this has happened before. And then okay. This is a little side note. What releases faster than, new models?

12:09 Somebody said it. That is amazing. JavaScript Okay. I'll shoot it all for you. I know you love dev jokes. Uh-huh.

12:22 But this has been your strategy. Let me hear you say, UPA. UPA. Say it like you believe it. Come on. UPA. Let me say, UPA. UPA. Alright. Don't say it too loud because it sounds like an agency that might get, you know. Some of y'all didn't like that joke. Alright. But here's the idea. I like that joke.

12:42 Alright. Understand, plan, and adapt. That is how you learned how to code. That is how you joined new companies. That is how you joined teams. That's how you built your start ups. You understood the problem. You made a plan. And then you adapted to the circumstances. Let me tell you what that means for you now. Understand.

12:59 First and foremost, we don't even have a good understanding of AI in general because we've been kind of avoiding it, which is normal, but we have a really good understanding.

13:14 This is time for you all to read that definition. Go ahead. Why don't I drink some water?

13:20 Artificial intelligence is a broad field and not a specific technology. Let me tell you what I mean. So there's artificial intelligence all the way at the top. Then you have machine learning, which is under artificial intelligence. But then you also have neural networks, which is under machine learning.

13:39 Then you have deep learning, which is under neural networks, which is under machine learning, which is under artificial intelligence. This is a broad space. This is not even the whole the whole stack. But what does this show you? That there is space for advancement, growth, career opportunities, problems to be solved.

13:54 There is space for expertise all through this. We see, we keep conflating generative AI as the end all be all. But it's more than that. And there's more for you.

14:07 You see, the field continues to evolve. A couple of years ago, all we cared about was the fact that we could generate images. But there are too many fingers. Right? We did a video of Will Smith eating spaghetti. It was a nightmare. Right? Still kind of scary.

14:21 And then we came up with this term AI slop because that made us feel better. Because as long as we could call it slop, then we didn't have to take it seriously.

14:31 I'm being honest. Just being honest. But now, what are we thinking about? Like, this is the year of agents. Right? And we're making up new words. Agentic. That is not a word.

14:43 But they keep saying it. And now it's what? MCP, model context protocol. How many people have seen at least five videos on their feed about MCP this week? Out of nowhere. And they just announced that in November. You see what I'm saying? Like, this is all new stuff. Everything is still developing.

15:01 So once you understand, you better stop. Quit playing with me, clicker.

15:09 Alright. See, here's what you have to do. I understand that you don't want to do this. I understand. But let me at least give you an idea. So if you learn to build AI powered apps and also use AI powered workflows, you get to do something that is not about you.

15:36 You gotta take yourself out of the whole picture just for a second. If you are able to leverage these incredibly powerful systems, imagine what you could build for your users.

15:50 None of us have a job in IT or technology that doesn't revolve around a user. None of us. Your hobby project can revolve around you. Sure. But guess who you are? The user. You see that? So the users are all users are the key. So think about what you could build.

16:09 You know, there are some very scary problems out there. Very scary problems. And I don't mean technical problems, I mean life problems that are beyond human scale right now, that we just cannot solve with people.

16:24 I want to put these things to work to help. I want a model to help cure cancer. You see? I want a model to help with Lou Gehrig's.

16:41 But who are the people who know what the problems are? Who knows where the needs are?

16:49 Are you are you starting to feel like wait. Is he talking about me? Because I am.

16:58 You are still the problem solvers. You are still the problem solver. So now you have to adapt problem solvers. You just have more powerful tools than you've ever had before at your disposal. Now you just have to adapt.

17:15 Change is the only thing that's constant now in our lives. We just have to accept that. We just have to accept. For a long time, for the last twenty years, JavaScript has been supreme and the only thing we all cared about. And now that's changing. Because you know what changed the most?

17:33 That the least interesting thing about you is your ability to code as an engineer. That one took me some time to just embrace that the least interesting thing that I can do is code now. It used to be a barrier for people because if you were the best engineer on your team, that gave you some status.

17:55 It gave you some power, some influence. But now the best engineer is the person who's using all the tools around them to solve the problems the most efficiently and the quickest.

18:09 Now those are the best engineers. But it doesn't mean your skills are are not valuable. Because guess what? This is why I say that you still matter in this whole game. Because you know what to do. You know what to do. So think about one of the demos that we saw today.

18:32 Devin came up here, and he made an app in twenty minutes and deployed it to the Internet. But why did that work? Because he knew what to do. He knew what the architecture was that he wanted to be a part of the application. And then he just accelerated the ability to bring that to fruition.

18:53 Is that the end of that app, though? It didn't have to be. It may be for today because it was a demo. But there is still more work to be done. There are still load balancing to happen. There are still security parts to be done. You see, but that's not about writing code.

19:06 It's about knowing what to do. You see that? It's about knowing what to do. Because those database queries that were generated work. But do they work at scale?

19:18 I'm not saying that they don't. But guess who knows if they do or not?

19:25 And see, you're still so valuable in this game. Because here's what I say your new value add is.

19:33 Learning architecture. If you are in here, you have not learned architecture in your career, today, enjoy the party and then tomorrow, start learning. Being dead serious. I'm trying to help you. Listen to me. Alright. I'm trying to help you learn about design patterns. Because you want to know, how do I know what to do?

19:54 How will I know? Yeah. I'm telling you right now. Architecture design passion. Guess what? As these systems start to be built and they grow and they spread, guess what's going to really be needed?

20:07 What does this represent, you think? Somebody said Go ahead. Say it loud. Integrations. Integrations. Knowing how to integrate systems that scale and that work and that serve who? Not your ego, but serve the users.

20:26 So I find it so interesting that we have all of this. You know what I mean? Like all of this is still waiting for us. It's just different. It's just different, man. They listen to me. They turn this thing off. I might be way over. Let me wrap up.

20:38 Let me wrap up because they just like, you know, we gave up on him. Here we go Keith.

20:49 I appreciate that. Don't worry. You all can still use the app to find a dinner partner. Don't worry. We got you. We'll have time for that. But another reason why I say the least interesting thing about you will be be able to write code because of this new word.

21:02 Imagine a world where your nontechnical collaborators and colleagues add features. They can't add my checkout cart. Okay. I'm not talking about that. But how many people here have built an app, and then they're like, can you move this button by five pixels? And you're like, do you know what I'm actually working on?

21:23 I'm actually trying to scale this algorithm so that it can, you know, get rid of the bottlenecks. You know what I mean? I got this new breakthrough. You don't want to spend your time moving pixels. Is that interesting? No. So just be prepared for the day when you're PMs. Just be like, okay. You know what?

21:41 I know you don't have time. Hey, Gemini, generate the code to move this button by five pixels. Here's a screenshot. Actually, while you're at it, open a PR for this. Run some integration tests. That's not a bad world. Because you didn't want to do it anyway. So don't even be mad. Why are you mad about this?

22:01 You didn't even want to do that work. So why even be mad that it can be so the chores can be given to something else and you can do the cool stuff that you actually care about. You see? Use A. I. To do the chores, the grunt work, the stuff that you really hate to do.

22:17 Some of y'all haven't written a unit test in six years. And I know it. I know y'all. Look, because I have I have it. So it's fine. It's fine. It's fine. It's fine. But the thing to think about is that you can finally stop saying, I'll write them in the next sprint. You could just do more.

22:38 AI algorithms, L. O. M. S, they're not perfect. You're gonna be needed. You know why? Because they're non deterministic.

22:48 Devin told you. During his talk, he said, you know how many times I had to try this? Because sometimes it just didn't work. But you know who you know who can tell that it doesn't work? Who can tell that when the quote that is generated is actually inaccurate?

23:06 Yeah, it's you. You know what I'm saying? Yeah, I think it's us. Yeah, it's you. Yeah, it's you. Alright. You gotta remember this is your strategy for the future is to understand, plan and adapt.

23:22 I know Listen, I'm not telling you not to be afraid. I'm just telling you, now you have a space to go after the fear.

23:29 That's all. There's a space to go after the fear. Because I felt it. I still sometimes, when people ask me at work, like, 'Hey, how can Gemini help you with this work?' I get a little tight. This is mine.

23:43 Because I lose sight of the fact that, well, if Gemini did this thing that could be automated, what else could I be freed up to do? What interesting thing could I be freed up to do? But that puts the onus on me because this work is defined. The, the, the stuff beyond.

23:59 That's where your engineering skills come back in for you, friends. So that third one, the next gen engineer who drives impact beyond code assist, well, they use all the tools available to create experiences that delight their users. That's who's going to win. That's who's going to win it all.

24:19 The people who can do that, they're going to win it all. The companies that have all engineers that can do that, they're going to win it all. You want to be on the right side of this. I'm not saying let the L. L. M. S do your work. I'm just saying don't be afraid.

24:35 Embrace, understand, plan, and adapt. Listen, we've been trying to build all these companies.

24:44 We're trying to build artificial intelligence. But if you have heard nothing else that I have said, if you heard nothing else, remember that you are valuable, that you are still needed, that you are a core part of this industry. And that while we're trying to build artificial intelligence,

25:03 you're already intelligent.