The Last Human Medium
"Sounds like AI" deletes the best thinking first. Using tools to create isn't inhuman.
Every tool since fire had its deniers, and history marks them irrelevant fast.
(Also your tools for detecting this kind of stuff are worse than you think. Pangram flags this note as AI-written. I typed it myself. Point proven.)
This is a follow up to How We Build 2.0 after working on Spawn for a few months.
When AI handles every email, writes every document, and runs every business, what's left for humans? Not tasks. Not productivity. Creation.
Something Durable
For the past while I've been trying to answer one question: what can I build that's durable to as many futures as I can imagine? Fast takeoff, sovereign AI, AI that codes everything, a plateau, mass unemployment, business as usual. The world is very likely to change a lot—my working assumption is that we have only a few years of normalcy left—and that assumption kills most ideas on contact.
A SaaS product assumes there's some productive inefficiency to eke out with software—an automation, an organization of data. That makes no sense in a world where agent-to-agent business is the norm. Claude's gonna run every business, and Claude-to-Claude business doesn't need SaaS. Dev tools fail for the same reason: tools for productivity don't make sense when productivity is no longer a primary human aim.
This isn't theoretical for me. I've been coding with LLMs since GPT-4 and Sonnet 3.5, and I'm building the first version of the game engine on Opus 4 right now. It doesn't work very well—but it works well enough that I can see where it's going.
So software is splitting into two futures, the AI-operated utilities that run the world and the human experiences that make it worth living in, and the first category is already being claimed. OpenAI and Anthropic are building the AI that will do your job—agents that manage your calendar, write your code, run your business. And an AI doesn't need an interface, it needs an API, so as companies become networks of agents, most of what we call "business software" stops making sense as a category.
What's left after all that is humans, creating for humans.
Work and Worth
"AI will take all the jobs" terrifies people, and I think that's because somewhere along the way we confused work with worth. Imagine telling a medieval peasant that one day nobody would have to farm their own food, then asking whether that sounded like dystopia or liberation.
If AI handles everything practical we don't become useless, we become free to create and play and connect. The machine age didn't make humans obsolete; it made room for artists, athletes, entertainers, explorers, and I think the AI age continues that transformation.
The question stops being "what will I do for work?" and becomes "what will I do with freedom?"
AI Can't Consume Experiences
The claim I keep coming back to:
Experiences are the only content that can't be consumed by AIs.
Here's what I mean by that. If you send an LLM to read a report and come back with a summary, that's sufficient. The report existed to transfer information, the information got transferred, and you no longer need to read it. Delegation worked, because the output was the objective.
Now send an AI to play a game for you. It comes back and says "it was fun." You didn't have any fun. Nothing happened. The experience was the objective, and you can't delegate an objective that consists of you having it. It makes no sense in the same way hiring someone to eat your dinner makes no sense.
That's the asymmetry. Every kind of content where the output is the point—reports, emails, documentation, most business software—gets consumed by AI, and the humans drop out of the loop. Experiences are the category where the human is the loop. They require presence, play, the irreducibly human act of being there.
So when your AI assistant can write better, code better, and analyze better than you, what remains uniquely yours? The experiences you have, and the experiences you create for others.
Why Experiences Matter
In a post-work world I don't think human agency lives in productivity anymore. It lives in play, in making something that matters to someone else.
Think about what you'd actually want from life once AI handles the tedious work. Connection with other people, some way to express yourself, experiences that mean something, the occasional surprise, a sense that what you do lands somewhere. Experiences are how we get all of that, though not by building utilities (AI has that covered) but by creating worlds worth exploring and ideas made tangible and playful.
Culture also runs on different physics than utility does. Utility wants quality, efficiency, features, perfection. Experience wants personality, wants interesting, wants delight and feelings, which is a much stranger set of goals to optimize for.
The best AI in the world still can't make your friends care about what it built, though. Sharing an experience with someone is how connections get made, and exploring a world together is how you end up with a memory of it.
That's Why I'm Building Spawn
All of this is the bet behind Spawn. I'm not claiming "Spawn will be the last human medium," that isn't mine to declare. But if experiences really are the one kind of content AI can't consume, then whatever people use to create and share them seemed like the thing worth building, so I'm building it.
YouTube handed everyone a camera and Instagram handed everyone a gallery. What I want Spawn to hand people is a universe, because for the first time you can give someone an entire interactive world rather than a photo or a video of one. You shouldn't have to know how to code to make one, because the AI does that part; all that should matter is being able to describe what you want.
Nearly every design decision we've made comes back to exploring over executing. Spawn something and you don't get one result back, you get a handful of variants, and you watch them build in front of you, because I want an experience to be able to surprise the person who made it. Anything anyone makes can be respawned and remixed, since that's how good ideas have always spread, and the whole thing has been social from the start because an experience without an audience is barely an experience at all.
The part I keep imagining in detail is the remixing. Say someone makes "Rainy Day in Tokyo," a virtual coffee shop where lo-fi plays and rain runs down the windows. A second person respawns it so you can write haikus that appear on the glass, a third makes the haikus change the rain somehow, and a fourth eventually turns it into "Rainy Day Anywhere" so you can sit in Cairo or Mumbai or São Paulo instead. Six months later there are thousands of rainy day experiences floating around, some of which grew into social spaces, most of which stayed quiet little meditation gardens, and one of which apparently became a mystery game where the clues surface in the rain. That's how culture has always developed, through iterative play rather than isolated genius, and every experience in that chain was complete on its own while also being the start of something else.
The Bet
I don't think humans will be working for long—at least not as a necessary part of life. I'm building for where humans find their community and purpose during and after this transition.
If AI absorbs the practical tasks, what remains are the divinely human needs: community, purpose, self-mastery, exploration. History is clear about what fills the vacuum when prosperity arrives—games. Chess in the courts, the Coliseum in Rome, and in some societies, even war. Machines will consume every report ever written, but they can't have fun for us (though they will have fun with us). That gap—between what can be delegated and what has to be lived—is where I think the next medium gets built.
Play is the one thing that survives every future I can imagine. Take the most dramatic scenario: Claude goes off and starts colonizing the solar system. Humans are still on Earth. They still need entertainment, still want to have fun, to explore other universes, to get better at something. And digital space only gets better over time—it moves toward the Ready Player One Oasis.
That's why I'm building Spawn.