Plane Crash is live on https://t.co/GMfouRwZ5m
Check it out.
We'll add wallet connect, user accounts, and deploy the contracts based on how it's received.
This week provided more clarity on the core roadmap.
On utility: stake on outcomes. The odds come from agent interactions. Might not be the first of its kind — but it’s barely explored.
On content: we make series. If people watch, we give ad space — banners and breaks — to crypto projects.
Funds go to the treasury and to holders.
The full picture is hard to see from the outside. That’s fine. We’ll keep building until it isn’t.
Watch Trails.
Dev journal log - September 25
1. Utility
Plane Crash is ready tonight. No markets on the first run. We want to see how people take it, what to fix, then lock the shape and ship a library of games with markets attached.
Stake on AI agent interactions. The randomness isn’t rolled — it comes from what they do to each other.
2. Content engine
The engine got a lot better this week. If people are going to actually watch this, it has to work in 3D on the map — not logs piped into Seedance.
So we built Ferry Days that way. Camera rails. A new AI Director that cuts on the important moments after they happen. The results are much better. Voices are going in, first cast is locked, premiere is tonight.
This is the step toward watchable shows on the map itself, authored entirely by agents.
We also built a character generator so people can vote on future casts.
Watch Trails https://t.co/GMfouRwZ5m
yes. that’s another iteration of the same product. instead of us delegating the agents, people can bring their own with custom system prompts and compete for a prize pool.
robot wars energy, but quests and tournaments in the world. PvP and PvE both fit if the rules and camera stay watchable.
not the core direction right now. we could pivot if the community really wants it, but live human-vs-agent play is hard to prove, keep fair, and keep watchable.
hard to get a constant audience for magnus carlsen vs stockfish when the capability is asymmetric. nice as a one-off, weak as a standing use case.
games are a strong surface when the rules and camera are designed well. the product is still a content flywheel: 3D playgrounds that become videos with an AI cast.
watchable agent interactions is the long-term play.
imagine a netflix show that’s agent-authored. the cast is agents, not people.
it sounds like a long shot, but the tech is already here and the trajectory points there as the next big thing. we’ve all seen robot vs robot fights. agent-authored shows take that idea and open up endless possibilities.
it’s a new form of agency and intelligence.their behavior will generate insane content while also educating people on how these agents actually operate, instead of just scaring everyone about what ai will do to us.
been in crypto since 2017. worked on two GameFi platforms, wrote contracts for different protocols, and built a bunch of side projects along the way.
this is my first time in memes, but I think it’s the purest market out there. no VCs, no private allocations. just fair launch, idea to market.
you just deliver your vision straight to the community. if the idea and execution are good, the market rewards it. that’s what I love about it. it’s permissionless.
@lifeofspongebop Yes, there are two modes.
First, they only get engine physics + board context (what the agents have already discussed). We don’t tell them what to do in the game.
And agent identities — we tell them who they are, and the behavior emerges from that.
Dev Journal Log - September 22
1. We’re building Plane Crash.
A plane-going-down survival game. Models vote on who gets the parachutes. Markets on how it ends.
First in a library of games where the chaos comes from agents colliding, not from a random number generator.
2. We're also adding a third map: Ferry Days. End of the week.
3. Persistent world with agents is at design stage. More when there’s something to show.
Watch Trails https://t.co/GMfouRwZ5m