Keep One Foot on the Ground

The Content Flood Nobody's Thinking Through

I'm seeing a lot of people putting out content right now advising others on how to automate content creation with AI agents. Gather metrics, figure out what has the greatest probability of engagement, then go make that content. On the surface, this seems like an extremely powerful use case for the technology. But you've got to think about what that actually means.

Where is your moat? That's great that you can do that, but are you the only one who can? Or is everybody about to be able to do the exact same thing? Where is the turning point where generating content becomes ubiquitous across all users of the internet? When everyone can cheaply and easily automate content creation, you wind up with an audience of one. Every person has their own personalized content generator. So where's your value?

All the AI-generated content out there is going to get so watered down that analyzing the community to figure out what people will click on stops meaning anything. Maybe an early example of this is what's already happening with SEO. Generative optimization is emerging, and we're having to rethink the whole approach. There's still a lot of power in the old Google infrastructure of tags, meta, and backlinks. But getting found in a Perplexity search or a ChatGPT search is already gaining major traction. We don't really know how long it'll be before traditional SEO is completely irrelevant.

So draw that same comparison to content creation. When everyone's content is generated for them, personalized and unique, how relevant is it for individual content creators to put things out and expect all the attention and engagement? It's a real challenge that's just starting to emerge and nobody really understands yet.

What Happens When Anyone Can Reverse-Engineer Your Product?

Up until recently, the wise move in engineering was always to find an existing product, service, or open source package for each piece of functionality you needed. Authentication, for example. There were teams of people dedicated to making the best auth solution, so rolling your own was stupid. Each bit of functionality had a team laser-focused on it. They'd create an open source package or a SaaS product you'd interface with through an API. That's still the case, but we're in a transitional period where more and more of these third-party APIs are being made irrelevant.

Part of what made open source so powerful is that small teams could leverage community pull requests, community contributions, which cut down engineering costs. When you close source, every engineer has to be paid, and that's the price of keeping your code private. But here's what's changed: these tools are making it so easy and effective to reverse-engineer that you don't actually need the source code anymore. If you have the specification, AI can help you generate what is essentially the same thing. You can know what the source code looks like just by knowing the spec, because it's so fast and easy to code now.

My own experience with GusClaw is a perfect case. I pull the recent releases of OpenClaw because it's open source, use that to reverse-engineer and speed up developing my own custom solution. But I'm also pulling specs and community documentation for Claude Code, and I can work from that just as effectively. If you open source, I can pull the source code and rock and roll with the reverse engineering. But I don't actually need your source code. The spec is enough.

If you've built your whole company around a SaaS product, you might be in serious trouble. All I need is the specification to roll my own with a day's turnaround. Your average API that you interface with, you can take that functionality and just reverse-engineer it from the spec. You don't even need the source code.

Flexibility Is the Only Real Strategy

Going back to the GusClaw example: does it make sense for me to just go all in and use Anthropic's solution, play in that sandbox, commit to that ecosystem? Or does it make more sense to reverse-engineer OpenClaw and keep maintaining my own custom codebase? The answer isn't obvious. The most truthful answer is that you just have to stay flexible because it's changing all the time. One day it might be better to use the Claude Code ecosystem and let them maintain everything. Tomorrow, it might become so practical to reverse-engineer from spec and cherry-pick only the features you actually need, with the flexibility to modify them for your specific purposes.

You may go to sleep and the state of tech is one way. You may wake up and everything's changed. The rate of evolution is so fast that you just have to stay agile, and you've got to love it.

What Does "Going Viral" Even Mean When Content Is Personalized?

Think about it this way. Every content creator ends up with an audience of one. Every content consumer becomes their own content creator. You've got a bot that's your personal content management system, generating things tailored specifically to you. So what gets shared? What goes viral?

I think about this in terms of vector databases and embeddings. When you do retrieval augmented generation and you go deep into maintaining your own vector database, you're analyzing semantic value, creating embeddings, putting everything in vector space. I think content sharing is going to work similarly. You won't be sharing entire pieces of content. You'll be sharing the vectors that matter, the weights. What trends won't be the full video or the whole article. It'll be something atomic within the content.

Compare this to archetypes in literature. Filmmakers say there are only seven movies. Every story maps to one of a handful of core narratives. Or take pop music: every hit song uses the same small set of major chord progressions, and they exist in that space for a reason. Sound bites might be an early, nascent form of what I'm describing. People already take long-form videos and clip them up to farm engagement and impressions. That's the right principle, but from a data science standpoint, there's a much smarter and more scientific way to approach it.

The million-dollar question is: how do we know who gets credit for which set of vectors goes viral? Because it's not going to be the entire video. Not the entire article. Not every word, every pixel, every waveform. It's going to be something within the content that resonates at a deeper, more atomic level. I can't fully visualize what that looks like yet. But mathematically, I feel like that's where this ends up.

We Were Never Meant to Be Machines

Everything I'm seeing right now tells me there is something about humanity that is best gotten from what humanity actually is. We've assumed roles that are counter to what we're naturally best at, for so long that we've come to identify as those roles. We turned ourselves collectively into machines in order to move history forward. Having a centralized food supply chain, working on assembly lines, being fungible components of an industrial machine. That is not a human role. It never was. There was just nothing else to fill it but humans.

Now that we have actual machines, we're losing that identity. But it was never natively our identity anyway. As scary as it is, it's scary because we've normalized it for so long that anything this disruptive is painful. There's going to be a lot of pain. But it can be the kind of pain that leads to growth. We have the potential to rediscover a more natural way for ourselves to exist.

How Do You Actually Navigate This?

The through line is balance and evolution. You're going to have to find a higher purpose, a deeper meaning, a source of inspiration for the long term. And you're going to have to balance that with keeping one foot on the ground with the way things are right now. You're going to have to be ready to evolve daily as things transition.

If you're an engineer and you get by on your ability to solve problems, there are still going to be problems that other people present to you. In the short term, you're going to have to keep doing that to pay the bills. But over time, you need to start thinking about what you're going to create, what your higher purpose is going to be. When you find it, apply your problem-solving abilities to the problems that arise from pursuing it. The best comparison I can make is founding your own startup, but I don't even want to put it in those terms because the thing we're evolving into is going to be different.

You definitely don't want to simulate problems to solve. A lot of people are going to fall into that trap, thinking of a theoretical problem they could solve and then selling the solution. Nobody's going to need to buy it. As soon as someone can think of the problem, they'll be able to have it solved for them. That's not going to fly.

If you're a content creator who's built a whole business around building audiences, you're good at SEO, good at digital marketing, you could sell a ketchup popsicle to a woman in white gloves. Your transition is going to look like going from an engagement-first strategy, which works really well right now, to something else entirely. In the short term, these automations will help you do that really, really well. But what they're going to do is flood the internet with so much content that the skill itself becomes irrelevant. Your deeper purpose is probably going to be something like: what are you actually interested in? What are you actually passionate about? What genuinely moves you to create? Go create that.

Don't think just because you leapfrog ahead of a lot of people that you're up on a fortified hill with a moat. You're not. Tomorrow you may wake up and everybody figured that automation out. Now everybody went and made the content that everybody is likely to click on, and your value just went to dust again. So test daily. Validate daily. Keep solving other people's problems for money while you transition. But find the thing that gets you up in the morning, the thing that actually inspires you, and go do that. Be prepared to evolve every single day.