Transcript
0:00 The problem with AI and jobs is that nobody knows what you and I are worth anymore. Talent Board is my answer to that. More on that in a minute. The standard career advice in 2026 is you got to build your portfolio, you've got to learn the tools, you've got to ship projects, and you've got to show not tell. And that advice isn't bad, but it leaves us all leaning into the one thing that AI is making 100% easier for everybody, and that is generating code.
0:24 Everything around you is accelerating because of code generation, and we see this in the actual larger macro era stats for our time. You see the number of code projects on GitHub exploding. You can see the number of apps in the Apple Store exploding. People are getting good at generating stuff. And when you can generate a lot of stuff, it sure is hard to prove that you know what you're doing because generation itself is essentially free. And I want to call out here, this is often portrayed as how can juniors get jobs, right? How can people early in their career get jobs?
1:00 And I agree who are early career feel this acutely. I know folks who are just getting out of college who are suffering through this daily. But really, it's a larger issue. The entire mechanism by which we prove we can do things is broken for everyone at every level. And this is true for the college grad who can't get hired as much as it's true for the mid-career PM who can't demonstrate what she built in the last year in a way that's explainable and understandable and truly shows her expertise. And the urgency is real because layoffs can continue to come in droves if you work in tech. Oracle started cutting up to 30,000 jobs recently. We have Block cutting 4,000 people. We have Amazon cutting 16,000 jobs back in January. We have Salesforce cutting thousands. We have Dell shedding 11,000. We now have over 60,000 confirmed job cuts for tech roles just in Q1. What's interesting about this is that at this point, these companies typically are not making cuts because they over-hired during the pandemic. That was happening in 2024.
1:59 That was happening in 2025. We're sort of past that now. The question now is companies are making assessments about what we are all worth at work and deciding this is how many people we need for the mission we have in front of us. And implicitly, they are attributing value to what we do with AI, right? Like how many people plus AI can we use to get the mission done? That's really the equation they have in their heads, and that of course begs the question, which people, why, which AI tooling, what is the role? All of those nasty details that go into figuring out how an integration actually works. In this world, all of us, and this is this is gut level, this is visceral, this is friends over dinner conversation, right?
2:39 5:00 p.m. we go to happy hour conversation. Everybody I know in tech is thinking about how we prove our value right now. How anybody proves our value, and that's not abstract. It is the most concrete professional question facing tech this year. And this is my attempt to offer you something constructive. Not a here's how to optimize your resume, not a here's what you can build in your portfolio. Those are all positive things. We probably need to be doing them, but instead, I want to give you a compass for orienting your work in a world where AI generation makes just about any kind of building thing free, functionally. And I want to give you a set of principles that you can use to navigate that world. Basically, to make your value visible. Now, just to give you the quick tour, the reason we care about production and the fact that AI does production well now is pretty simple. Production used to be hard. Hard signified effort. Effort signified expertise. If you could do the product well, all of that added up to I know what you can do and I know what you're worth. All of that is breaking down now.
3:40 If you can make something that looks really good at first glance with almost no effort, none of the rest of that chain of value holds, and that's why we're all at 5:00 p.m. at happy hour wondering what's next, who's next, where the next layoff is coming. That's the core issue we're all facing. And it's not just a hiring problem. It's effectively a talent allocation problem for all of society. How does a company know who to promote? How does a team know who's actually contributing? At an even larger scale, how does an economy route talent to the work that matters the most? All of those questions used to be very easily answerable by this chain of value, and they're not anymore. So, here's what I want to offer instead.
4:17 Five principles that orient your work and together give you a compass to navigate this world and prove your value. Before I get into the framework, quick thing, everything I'm about to walk you through, it only works if your proof of work lives somewhere visible. Right now, most of your AI work is scattered, expired lovable URLs, clawed artifacts and dead conversations, repos nobody will find. So, I built Talent Board to fix that. Public profile your projects, prove you actually understood what you built. I'll come back to this at the end. But if you want to get ahead of it, the link is on my Substack. Okay, five principles. Principle number one. I think this is really important for 2026.
4:54 Think about comprehension more than you think about generation. The default instinct when you get access to super powerful AI tools is to build and generate a lot, to produce, right? Build stuff, ship stuff, generate output. This is the path of least resistance, and increasingly, you're optimizing for the wrong thing. And I don't say that cuz I don't love shipping. I love shipping. I love shipping fast. I love that AI lets us do that. That's great. But a deeper optimization here is to understand what you produce at a very fundamental level in context, to the point where you can explain not just what does this code do, but why does it work that way? What would break if I changed it? What tradeoffs did I make along the way? And what did I deliberately choose not to build? If this sounds obvious to you, congratulations. It actually isn't.
5:41 Watch what happens when most people vibe code a project. They will prompt, they will iterate, they will get something working, and they will ship it. At no point do most people stop and build a mental model of what is really going on with the codebase. In fact, that is one of the strengths of these tools. That is something lovable deliberately obscures when they give you the option to type something and get magic out. It's a magic box for an app, and they sell on that. And so do the dozens of other companies in their vertical. And if you asked most people using those tools why they used AI to make the choices they made, most of them would shrug and say, "It's what I wanted. It's what I dreamed. It's what I built. It just works, I guess." Now, multiply that by every person shipping AI-assisted work in every company in the world. The industry is producing at unprecedented speed and comprehending at unprecedented lows, right? Teams are going to deploy features these days that nobody on the team fully understands. PMs are shipping prototypes they can't fully explain.
6:36 Engineers are merging code that they can't fully hold in their heads. And the gap between what software does and what anyone thinks it does just keeps widening because we keep generating more of it. And it's getting people into real trouble. Amazon learned this very expensively when an engineer used their mandated AI coding tool, and the tool decided the optimal path was to delete the entire production environment. 13 hours of downtime for AWS, and the official response called it user error.
7:02 The user was following a corporate mandate to use AI tooling. This is what happens when production outruns comprehension at an organizational level. At an individual level, it can look less dramatic, but the failure mode is often the same. You're building things that you do not really understand, and eventually, something is going to break that you can't really diagnose. And so, what does comprehension mean? You have to force yourself to understand at the point of creation, to think carefully before you ship. Sit with it and answer the question, "What does this do? What does it do for customers? What are the dependencies here? What is the blast radius if something goes wrong with this piece of code? What decisions did I make and why? Where is the AI's work that I am choosing to throw out and the AI's work that I am choosing to keep? When did I last override AI?" This is not navel-gazing. This is not an exercise in documentation for its own sake. It is the practice of understanding your own work, and it is harder now because there is so much generative noise out there.
8:01 And the practice, that's the seed of what we call taste, right? Taste doesn't come from a mysterious aesthetic instinct. It's not like we all wake up one day and one of us is blessed with being Jony Ive for the day, and off we go. It comes from having understood enough things deeply enough that you start to recognize patterns, right? What works, what survives, what really matters. Comprehension is that pathway to taste, and taste is a skill that helps you survive in this world of change. It's poorly taught, but I think it's fair to call it a skill. Now, if you're getting started in your career, this is becoming the way you replace apprenticeship, right? The old way was often years of supervised grunt work as maybe an L4 engineer or junior PM, right? Ticket triage, documentation updates, test coverage, data cleanup, the stuff nobody wanted to do. But doing it was the education. You absorbed context. You learned what good looked like by seeing it from other people. You built taste through a thousand small tasks. That whole process is disappearing, or in a few cases, companies are doubling down on it, which I think is very smart. But the interesting thing is, in the age of AI, the learning mechanism underneath that apprenticeship does not take a corporate structure. You can build it yourself.
9:15 The reps come from forcing yourself to deeply understand everything you produce. The volume here matters less than the depth, and I see a lot of people optimizing for volume, and there are roles that care for that, and that's great, and you can get there. It's easy enough to push that gas pedal with AI. But one project that you fully comprehend teaches you more than 10 projects that you vibe coded and didn't think about. The AI will accelerate your production. You have to deliberately decelerate to make sure you understand enough that you can eventually go quickly with good taste. And I have seen senior engineers, I've seen other people who are deeply fluent in their domains go extremely quickly with AI once they unlock here. But if you don't take this time, you're just going quickly without knowing how to steer the car, and that usually results in a car wreck.
10:02 Principle number two, we need to make the ability to explain something clearly and very concisely its own class of artifacts. If we talk about comprehension as an inner task, right? This is something we do to like grow in our ability to develop as AI native workers, that's fine. How do you show it? Explanation is how you show it, right? is a scarce skill, explanation is how you make that scarce skill visible. It's not a blog post about your journey, it's not a case study you write after the fact. I've seen all of those, right?
10:35 It's actually a structured explanation that travels with the work itself. It's actually inseparable from it, right? As native to the deliverable as code is itself. This sounds pie-in-the-sky. What does it look like? I'm asking you to think about shipping with everything you make, what is this? Not a marketing description, just a clear plain English statement of what this does and what it doesn't do. Why did you choose this, right? What alternatives existed? What tradeoffs did you evaluate? Show what you made choices about where the choices were hard. What's going to break? Where are the fragile points? What assumptions is this built on? If the requirements changed, what's going to fall apart?
11:13 This is your blast radius question. It's the one that separates people who understand systems from people who don't. Answer it. What did I learn, right? Not in a self-help sense, concretely what did you discover during the process that changed how you approach something? Where did the AI give you something confidently wrong that you corrected? What would you do differently next time? I'll tell you one that I learned recently. One of my biggest learnings in 2026 is we the community have worked on the open brain project has been understanding the importance of typed definitions and schemas to making sure we can build things that scale. I knew that conceptually. I could tell you that I understood what a schema was. But seeing it come to life in the community and seeing where the schema works and doesn't work and how we evolve it and how important those choices are, I have a much more visceral sense of that now.
12:00 That's an example of something you can learn by building. And yeah, this is not a doc, this is not a spec, it's deliberately very simple. That's it, that's all the questions. The point is not to give you a lot of work, the point is not to give the reader a lot of work. The point is actually to communicate a degree of comprehension. And people will tell me, "Well, Nate, we can short circuit this, right? Nate, I can have Claude just do this for me." You can, but you're going to get caught if any human talks to you because they're going to see right away that it's a bunch of slop and you don't really understand.
12:29 Think of it this way, the explanation artifact in the generative era is essentially what the commit message was in the traditional software engineering era. A commit without a message is technically complete, but a commit with a thoughtful message signals that someone understood what they were changing and why. Same idea. Let's resurrect commit messages. Principle number three, think about transactions over credentials. Fundamentally, credentials are becoming stale. It is easy to tell ChatGPT, "Make me a senior thesis." I know people who do. This is inflating away the value of a lot of the things that we attributed value to in the past. Everybody knows that. But I want you to look deeper at the career arc. Look at the history of how your job unfolds over time as a series of what?
13:14 Transactions. That's right, you are transacting labor and receiving income. That is a deal you are making in the marketplace. And so really when we look for proof of value in someone's resume, when we say depending on experience, whatever we do when we talk about the value of a track record, what we're really fundamentally saying is transactions matter because transactions are where real money changes hands for real labor and real value. What I am saying is that in the world of AI, we have a velocity issue with that loop. Because when we were doing work before, it did take two years, three years, four years to do meaningful work. But now with AI, you can do meaningful work really, really fast. And yet the old way of looking at transactions, multiple years, how long have you spent, just hasn't caught up, right? Those career trajectories don't matter. And so we need to find a way to showcase real meaningful work that represents transacted value that may have happened in a very compressed period of time. Effectively, we need microtransactions for jobs. And that's something that I'm playing with an experiment on on the community side with the Substack group, and I'll talk to you more about that in a couple minutes here. But I think it's a larger issue. I want to talk about the idea that this is a fundamental new unit of value that we will need to replace resumes. Because you can't substitute for I did the work and I got paid for it. That's a real signal in the marketplace, and we should have a much richer history of those signals than just a job once every two years. In the age of AI when we're moving fast, we need micro job transactions. It's really important, and I want to look as a community at how we can build and expand on that more in that a second. First, principle number four, we're going to go through these five principles and I'll tell you what I'm building. Principle number four, work in the open. Not super surprising, but it's really important. If you're not working in public, you're missing out on a chance to be observed, right? I know that professional development used to happen behind closed doors. You would build your skills in private inside a company, you would be observed by a small number of colleagues, nobody outside your org saw the reps, but it didn't really matter because those people who saw you could reward you.
15:23 That whole system, as we've been discussing, is kind of in jeopardy right now. Part of the problem is that this model requires access. You need to be inside a company, and if you're getting out of college and you're looking for a role, you're not inside a company. If you're just fired or laid off, you're not inside a company. And if you are inside a company, they may not be looking at those signals anymore. So in that world, working in the open is a way that you can show your skills to the world. In a sense, this is taking the idea of a side gig and putting it on steroids. How do you show your value to the world publicly when you are working on something? It's sort of like the same idea that Venmo had, where Venmo made payments social. Venmo took these little transactions for dinner with your friends and turned it into a game where you can see what were your friends doing, what were the cute emojis they were sending to their other friend, etc.
16:10 And they made payments feel social, and that helped them to incentivize their app. We sort of need that open-source social ethos for our work. I know that we have GitHub for PRs, and for engineering folks, something like GitHub makes a lot of sense. But we as a community now all have access to these generative tools. So we all need something like this to show the work that we're doing in approvable way. And I think that that's super important, and something that I've been working on building. And look, I get it, working in the open creates accountability, it's uncomfortable. You might wonder, is my boss going to see me? Is that going to increase the odds of a layoff? I know it's 2026 and you should be able to tell your boss, "Look, I'm working in the open because it's good for my professional development." I know you can't always do that. I'm not trying to say in every situation there's always one-size-fits-all.
16:58 I am trying to say in general, working in public is a better odds bet than it has been for the last 20 years, and it's probably better than working in private is at this point. I know it's uncomfortable, I know people are going to be looking at you, and that's kind of the point, and it's okay. Principle number five is really simple and really fast. You know how I talked about we need to have sort of these commit messages for stuff, you got to ship that with the work. If you want to have a living resume, if you want to have something that goes beyond the resume, something that goes beyond a profile page, that's something that shows what you are worth. Remember the top of this video, we talked about how you know what you're worth, you have to ship your proof, the proof that you thought about this with the work itself. It can't be separated because then someone doesn't know if you worked on it or if the AI just faked it. And if you can separate out those two pieces, it basically becomes an invitation to spam. And now I come to the thing that I've been building for the community. I want a place where we can take the work we've been doing, and we can show it. And I don't just mean GitHub. GitHub is great, I love GitHub, sharing GitHub links on your profile is welcome. We put up a talent board, and we encourage people to share GitHub links as a part of that application process, etc. Fine.
18:06 Fantastic. GitHub alone is not the answer here. You need a place for the glorious mess of all of the AI artifacts you're putting together, and a designated place to say, "I can still think. I haven't lost my brain. This is how I approach tradeoffs. In fact, this is how I deepen my understanding by building this stuff. I can prove that my worth is increasing because my comprehension is increasing, because my skill set is increasing, because I can demonstrate that I understand from these projects what I can do." And so we're enriching the profiles on talent board to deal with this, and we're making them easier to share. Because what it means to prove your worth in 2026 is increasingly tied up in the Claude artifact or this weird press release page for something you built for your company and now you can't find it. Or maybe it's in a lovable that you built.
18:52 You need to have a home for that where you can say in a visually attractive way, "This is what I've been working on." And critically, you can't separate it from this is how I think about it. This is how I understand it. This is how I demonstrate that I know my worth and know how to make AI work for me in a way that preserves human intent, human knowledge, human understanding. Because if you can't communicate that part, everyone's going to assume you're a slop factory, and you don't want to be that person in 2026. And so that's what I built. I want to make it as easy as possible for you to put together a profile, to share it, and I don't want to be the person who gatekeeps all of this. If this sounds like a good idea to you and you're like, "Nate, I want to have nothing to do with your Nate network. I just want to put this up and take your principles and put them on my personal website, or I want to compete with you. I want to start a different talent network." Fantastic.
19:40 I don't care. I love it. Go to it. You know why? We need more of this answer in the world. I don't care about my network in particular, I'm just trying to experiment in public with a solution to this problem. So if you have a better idea, go build it. And I love that. Go show how we can build something that demonstrates our worth in 2026. Because it is one of the biggest questions for workers all over the world. It is what we are struggling with at happy hour.
20:04 So, this is your compass. This is how you need to think about proving your worth in a world that is going to tempt you by saying just 100x your output. That is only partly true. You do need to increase your output in the age of AI. I don't want to hide the ball on that. But, you also need to show you can think, prove you can think, prove you did work, and make that all shareable. And that's what I put together, and that's what I invite you to wrestle with, put together, join me on Nate's Network, or whatever it takes for you.
20:31 But, know your worth, and know how to demonstrate it in 2026. And that really does require cutting across the grain of the popular narrative, which tells you to leverage these AI tools, right? It's It's very easy to look at the front pages of Lovable, look at the front pages of ChatGPT, of Claude, and say, "I do want to build something. I do want to think. I do want to maximize my potential by generating stuff." Fantastic. Remember to prove that you can think. Because that is the rare commodity now. Best of luck, and know your worth.
21:01 Everything I just walked you through, it only works if it lives somewhere people can actually find it. Right now, your AI work is scattered. Expired URLs, dead conversations, repos nobody will find, no context, no signal. That's what Talent Board is for. Public profile your projects, the four questions answered, proof that you can think, not just generate. You can find the link in the description below. I'll see you there.