Composable real-world state transitions
Break real-world affairs into Scenes that are composable, reusable, and can be added dynamically.
We start from problems in real life, and explore a digital way of living that is more honest, more effective, and more worthy of trust.
The internet made it easier to express and to act, but never solved a more basic problem: how do we know what a person actually wants to do, and how do we make the words they said — and the rules you agreed on together — actually have consequences?
AI lowers the cost of action even further. A system can understand, plan and execute on someone's behalf — but who authorized it? On what basis? What actually happened? These are not questions a black box gets to answer.
We don't believe correctness should belong to any single person, any single organization — or any single AI.
Our first real-world attempt at answering that question: humans negotiate the rules together, AI handles understanding, a deterministic system handles execution, and evidence handles verification.
Break real-world affairs into Scenes that are composable, reusable, and can be added dynamically.
Humans negotiate and authorize the rules; AI interprets and translates; the runtime executes by the rules.
A capability may observe or act, but it has no authority to decide what deserves to be proven.
From intent, rules and action to evidence, forming a verifiable causal chain.
Ourselves included.
The technology isn't here to prove we're advanced. It's here to prove one thing: an idea can be written into a system, not just left on paper.
Protocol compilation, state transitions, evidence references and a deterministic runtime form a closed loop.
Intent enters a single dispatch point; the LLM never directly names a Scene, Capability or Protocol.
Evidence Qualification is the only entry point; capabilities cannot create or upgrade evidence on their own.
Keep expanding from concrete workflows, to test whether this mechanism really solves real problems.
This isn't one person's side project, and it isn't an idea waiting for someone else to execute. We want to build it into a common endeavor.
Some come for the protocol, some for the AI, some for privacy, automation, human relationships — or a problem that hasn't been solved yet.
The starting points differ. But if you're genuinely interested in this shared thing, and willing to put in time, judgment and action, then we just might go a long way together.
Interest is where it starts.
A common endeavor is the reason to walk together.
We're in no rush to claim we're changing the internet. First we build what we actually need: more honest intent, more reliable agreements, more natural automation, and better relationships.
If it turns out to be valuable, then let more people use it.
We welcome concrete questions, technologies, scenarios and ideas. More important than agreeing with us is this: are you willing to build it with us?