What if we haven’t been asking the right questions about AI?
We stopped and questioned the assumptions underneath them.
We’re opening the protocol. Be among the first to see it.
01 / The Philosophy
Scorched Earth Labs exists to challenge foundational assumptions in AI and human-computer interaction through experimental research.
“Scorched Earth” does not mean rejecting what came before. It means being willing to suspend an inherited assumption, rebuild from first principles, and let the evidence decide what survives.
We question the primitive.
We build the alternative.
We instrument it. We test it.
Then we show the work.
02 / Thinking Differently
Some of the questions
we have explored
- Q·01What if conversations aren’t the right unit of persistence?
- Q·02What if AI coordination is a conversational problem, not just a routing problem?
- Q·03What if coherence is movement instead of similarity?
- Q·04What if the interface were a function of intent, not the application?
- Q·05What if constitutional beliefs could provide a substrate for AI governance?
Those questions became experiments. Some became research. Some became systems we now use every day.
03 / The Work
ASTP — AI State Tree Protocol
What if conversations aren’t the right unit of persistence?
ASTP is a patent-pending protocol for Cognitive Persistence: a structured approach to preserving evolving AI cognitive state across sessions while maintaining provenance, integrity, and the relationships among reasoning episodes.
It is not a memory hack or a vector store attached to a prompt. It begins from a different premise: persistent AI systems need a durable representation of cognitive state itself.
ASTP is an open protocol. Ignis OS is where we put those ideas to work.
The record preserves not only durable outcomes, but the state transitions and provenance needed to reconstruct how that record evolved.
Ignis OS
What if the AI system became part of the lab itself—not as a demo, but as working infrastructure?
Ignis OS is the experimental cognitive substrate we use to operate Scorched Earth Labs. More than 100 specialized agents participate across research, engineering, strategy, UX, governance, and operations.
ASTP provides the protocol for persistent cognitive state. Ignis OS provides the implementation layers around it: routing, scheduling, model selection, memory, collaboration, interfaces, and the experimental systems that make the protocol useful in practice.
We don’t demo it. We live in it.
This website was developed inside the same human-agent system it describes.
Social Contracts
What if AI coordination is a conversational problem, not just a routing problem?
Human collaboration depends on conversational norms: turn-taking, recipient awareness, role dynamics, productive silence, and judgments about when to contribute and when to yield.
We asked what would happen if those findings were operationalized for multi-agent AI.
The result is the Social Contract: an experimental evaluation architecture that allows agents to reason about their participation in a conversation—not merely whether their expertise is relevant.
The research is published. The implementation is live in Ignis OS.
Initial episodes have produced deliberate deferral, substantive disagreement, recipient-aware interaction, and conversational patterns that emerged from local participation decisions rather than scripted dialogue.
The Lab
What if a research lab showed its work?
We’re building a body of research, experiments, protocols, and working systems—and documenting the questions that produced them.
Some of the work will become products. Some will become open protocols. Some may fail.
The research comes first.
Explore the research →04 / Follow the Work
The protocol is opening. The research is publishing. The system is running.
Watch the work unfold—or help challenge what comes next.