Today on the AI Daily Brief, why agency is the most important attribute for the next decade. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes.
Hello, friends. Happy weekend. It being the weekend, we are back with another Longreads episode. And this one got a lot of discussion going on AI Twitter this week. It's called Agency is Eating the World. It was written by Gian Segato, a founding data scientist and engineer at Replit. Now, Replit is, of course, a vibe coding platform. And so Gian is living right in the middle of these transitions.
Today, we are going to once again turn this over to an AI version of myself from 11 Labs to read John's piece, and then we'll come back and discuss it with the non-AI me. In 2023, Sam Altman famously said, there'll soon be a one-person billion-dollar company. Two years later, we're watching his prediction unfold, and not simply because of AI, but because of the individuals who are wielding it. A new breed of companies is emerging, lean, unconventional, and wildly successful.
They generate hundreds of millions of dollars yet have no sales teams, no marketing departments, no formal HR, not even vertically specialized engineers. They're led by a handful of people doing the work of hundreds, leveraging machines to scale their impact. For years, we feared automation would replace humans. But as AI reshapes the economy, it's becoming clear that far from replacing human ingenuity, AI has amplified it.
The critical dividing line in our economy is no longer simply education or specialization, but rather agency itself, the raw determination to make things happen without waiting for permission. Agency. The word agency is thrown around a lot these days. It's not always clear what it means. Many people in tech today, myself included, are working to create digital systems capable of interacting with and adapting to the external environment to achieve specific outcomes.
We call these programs agents. A legal agent might autonomously scan court documents and compile a legal strategy. A financial agent might continuously monitor market conditions and dynamically adjust pricing or investment strategies.
For the past three years, I've been part of the core team behind Reaplet, a coding agent that writes, runs, and deploys software programs autonomously. Despite its status as standard industry nomenclature, I've grown to dislike the term "agent." I find it inaccurate. When building agents, we're giving programs everything except "agency." The appeal of these tools lies in their combination of capability, maneuverability, and predictable behavior.
They execute complex tasks while remaining fundamentally responsive to instructions. When some time ago, a few companies started offering more independent coding models, customers didn't like them. The most successful AI products today are reactive, not proactive, and don't exhibit genuine independence, primarily because that's what customers want. Programs that listen when told what to do.
High agency people don't. True agency is an unruly psychological trait. It's the willingness to act without explicit validation, instruction, or even permission. It's the meme, you can just do things, knowing that you can poke life and something will pop out the other side.
It's a venture capitalist with no academic background, founding the most important AI lab in history, a gaming entrepreneur creating a $30 billion military company, a fintech CEO birthing the private space industry out of thin air. True agency involves defiance, improvisation, instinct, and often irrationality.
You don't need to be in tech to have high agency. Yet you'll find most of these people there, as they naturally gravitate toward low-structure, high-impact environments, namely, startups. While driven individuals capable of making things happen have always existed, their bandwidth has historically been limited.
It's impossible to scale impact entirely alone. We live in a complex world. To accomplish anything significant, you still need to specialize at some point, which takes time. High-agency people aiming to achieve their goals still need to hire specialists or spend years selecting and mastering a discipline. When I founded my previous company a decade ago, it took me nine months to go from idea to working prototype. Nine months spent essentially learning the basics required to build a digital product.
And yet, this effort still didn't put me anywhere near the proficiency of a professional software developer, designer, or marketer. Becoming an expert takes time. Specialization is a necessary precondition for success, making it a form of local monopoly.
We live in a homeostatic equilibrium. Even the most intelligent and driven people can't possibly perform every job simultaneously. Stable ecological systems sustain themselves long-term precisely because the apex predator can't be everywhere at once. This dynamic explains why being more skilled and educated confers a persistent advantage, and why on average our society favors credentials over outcomes. The optimal strategy is typically to follow the lead,
stay in your lane, do your homework, and achieve top marks. To get a job, keep that job, specialize, and climb the ladder. There's a reason college tuition has outpaced inflation multiple times over. People hope it will help them develop specialized skills, creating their own form of vertical monopoly within the workforce. We don't live in a world that's kind to generalists. Well, until AI, that is. A phase shift.
We're now facing a rupture, a phase transition. AI has eroded the value of specialization because, for many tasks, achieving the outcome of several years of experience now takes a $20 ChatGPT subscription. If a decade ago it took me nine months to gain enough experience to ship a single prototype, now it takes just one week to build a state-of-the-art platform ready to be shipped.
a project once only achievable by a full team of professionals. Skeptics will push back against this techno-optimism, arguing that AI is sloppy, intrinsically probabilistic, and prone to mistakes. Build using AI, the story goes, and things will sure enough eventually break. You need an expert to know when to trust the machine. I understand, but disagree. While it's true that specialization hasn't become irrelevant, it no longer matters indiscriminately and uniformly.
I expect a bimodal-shaped distribution of AI deployment, depending on how comfortable we are as a society and with risk. In industries where untrained people equipped with imperfect AI models could make costly mistakes, we are going to see demand for specialized human accountability. This will include sectors such as defense, healthcare, space exploration, biological research, and
and AI advancement itself, all domains where the variance of prediction models is higher than the acceptable risk threshold. Wherever mistakes can kill and AI can't prove to be virtually all-knowing, we can expect regulation to enforce natural barriers and the need to hire experts. It's similar to why we continue to require human pilots despite having the technological capacity for autonomous flight. Sometimes we just want the ability to point a finger.
However, for most jobs, this is not true. Wherever we are okay with trying again after getting a bad AI generation, we will see market disruption. Data science, marketing, financial modeling, education, graphic design, counseling, and architecture will all experience an influx of non-specialized, high-agency individuals. Sure, machines will keep making mistakes, but their rate of improvement has been astronomical and will only continue to delay the point at which generalists feel the need to hire experts.
Three years ago, AI could simply auto-complete small code snippets. Two years ago, it started fixing broken programs. Last year, it began creating new projects from scratch. It now can autonomously understand large projects created by thousands of human developers and be used even by non-professionals. The game has shifted, and the winning strategy with it. It's no longer about understanding specialized details. It's about grasping the high-level global picture.
It's less about knowing how to patch a system and more about knowing that it needs to be patched. It's more about architecture and less about implementation, precisely where generalists thrive. For these individuals, boundaries between professions are beginning to blur and overlap. I've begun to see product managers developing business financial models, designers writing commercial ads, barbershops building custom booking systems, and restaurant owners creating advanced pricing tools.
Even domains seemingly far from tech, like agriculture, are beginning to see this impact, with farmers building crop tracking systems. These people always had it in them to do these things. The key difference is that it no longer takes years to learn how.
If you extend this argument to the limit, you end up with individuals running entire companies by themselves. The share of solo founder startups has almost doubled in the last few years, and the first examples of businesses with a handful of employees generating hundreds of millions of dollars in revenue have emerged. Henri Shi, who spent nearly a decade growing Super.com into a $150 million ARR business, is now tracking his progress towards Altman's one-person billion-dollar company goal on a public leaderboard.
He reports an average of $2.8 million in revenue per employee, coincidentally the same as Apple, the most valuable publicly traded U.S. company of the past two decades. Companies like Midjourney, with its 40 employees and $500 million of yearly revenues, represent a structural shift, not an anomaly. These companies are now full of high-agency individuals carrying the work of several teams.
and are easily competing with much larger companies. This is the unraveling of credentialism. Having an edge in the market is no longer about knowing how to do something very specific very well. It's about being biased toward making it happen. A new world. My entire world model has collapsed into a single bit. Agency or no agency. The transition will take time, and it will be far from painless. Institutions built around credentials won't go gently into the good night.
Middle management will fight to keep headcount, because it will take time to shift away from the idea that more people working on a problem signals a more important problem. Schools and colleges will take a while to adapt their teaching methods and content. Only bottom-up market competition will force change,
Structure isn't always bad. One-person companies come with high chaos potential, and cascading errors can be hard to contain at scale. A solo operator lacks the redundancy of a team. When the AI fumbles, there's no safety net, and small mistakes like mishandled tax compliance can snowball into audits, bug fixes, social media rants, and refund chaos. Yet these entrepreneurs will still prove to be mighty competitive forces that big firms can't ignore.
The good news is that having high agency is an internal state of mind, and it can be absorbed. It means freeing oneself from artificial constraints and defying them. It's Morpheus challenging Neo in the Matrix. Do you think that's air you're breathing now? The limitations we've accepted as natural—degrees, credentials, specialized skills, years of experience—are no longer the barriers we believed they were to making things happen.
Like Neo, the hardest part is simply believing we're free to jump. Today's episode is brought to you by Vanta. Vanta is a trust management platform that helps businesses automate security and compliance, enabling them to demonstrate strong security practices and scale. In today's business landscape, businesses can't just claim security, they have to prove it.
Achieving compliance with a framework like SOC 2, ISO 27001, HIPAA, GDPR, and more is how businesses can demonstrate strong security practices. And we see how much this matters every time we connect enterprises with agent services providers at Superintelligent. Many of these compliance frameworks are simply not negotiable for enterprises.
The problem is that navigating security and compliance is time-consuming and complicated. It can take months of work and use up valuable time and resources. Vanta makes it easy and faster by automating compliance across 35+ frameworks. It gets you audit-ready in weeks instead of months and saves you up to 85% of associated costs. In fact, a recent IDC whitepaper found that Vanta customers achieved $535,000 per year in benefits, and the platform pays for itself in just three months.
The proof is in the numbers. More than 10,000 global companies trust Vanta, including Atlassian, Quora, and more. For a limited time, listeners get $1,000 off at vanta.com slash nlw. That's v-a-n-t-a dot com slash nlw for $1,000 off.
Today's episode is brought to you by Superintelligent, and I am very excited today to tell you about our consultant partner program. The new Superintelligent is a platform that helps enterprises figure out which agents to adopt, and then with our marketplace, go and find the partners that can help them actually build, buy, customize, and deploy those agents.
At the key of that experience is what we call our agent readiness audits. We deploy a set of voice agents which can interview people across your team to uncover where agents are going to be most effective in driving real business value. From there, we make a set of recommendations which can turn into RFPs on the marketplace or other sort of change management activities that help get you ready for the new agent-powered economy. We are finding a ton of success right now with consultants bringing the agent readiness audits to their client
is a way to help them move down the funnel towards agent deployments, with the consultant playing the role of helping their client hone in on the right opportunities based on what we've recommended and helping manage the partner selection process. Basically, the audits are dramatically reducing the time to discovery for our consulting partners, and that's something we're really excited to see. If you run a firm and have clients who might be a good fit for the agent readiness audit,
reach out to agent at bsuper.ai with consultant in the title, and we'll get right back to you with more on the consultant partner program. Again, that's agent at bsuper.ai and put the word consultant in the subject line. All right, back to real NLW. A couple of different points of conversation. I think it's a great piece. It's very thought provoking, and I'm really excited that Gian has shared it and gotten us all to chatter.
Now, agency is a safe bet for the term of the moment when it comes to how people are thinking about the transformation we're in on a very high and very broad level.
You've probably heard people like Y Combinator's Gary Tan talk about agency as the most important skill. And in many ways, this is sort of becoming conventional wisdom in Silicon Valley, although that's far from the case outside of it. There are actually a number of different points that Jian is making here. One of them is about the erosion of the value of specialization. I think Jian's division of the world into effectively risk profiles were the
where the consequence of errors on the one hand are extremely high, or the consequence of errors are on the other hand fairly low, kind of makes sense and I think is fairly likely to come to fruition. In fact, we're even seeing this sort of division in the way that companies are experimenting with agents right now. There are certain parts of their business where they simply can't abide the current fail rate or hallucination rate or underperformance rate or however you want to determine it of agents because it's so mission critical.
On the other hand, there are areas where the consequence of those problems is simply less pertinent. It is in those consequence-light areas that companies are getting their reps with agents now, with the knowledge that capabilities continue to trend up. The other part of this argument, however, is about how much more individual people can do. Or in other words, how much farther their own agency gets them in a world where they can also hire and deploy AI.
The point that I wanted to make here is actually less about startups, which are Gian's focus, and more I wanted to connect the dots back to the Microsoft Work Trend Index that we talked about earlier in the week that basically predicted that the end state of agents in the office is human orchestrators and agent operators. Basically that humans were going to do the planning and that agents were going to do the execution.
That's a different way of saying that the key skill sets and attributes of people in the workforce in the future is going to be around planning and coordination of agent armies or swarms or whichever term becomes popular. And this is why I think the current crop of upskilling and AI education platforms is so fundamentally off. They're still operating on a model where the skills that matter are things like prompting co-pilots and assistant tools, when the reality is that everyone has to turn themselves into agent managers or, to use Microsoft's term, an agent boss.
I think that the more that we recognize and prioritize the sort of agency that Gian is talking about here, the more quickly we'll be able to reprioritize how we think about training, skills, personal development, and more as we head into this new agentic era. Anyways, one more big thanks to Gian for writing this piece. And thanks to you guys, as always, for listening. Until next time, peace.