VANTA world runtime
Native world, rendering, simulation, persistence, and machine-learning seams organized around host-owned state.
Produced byVANTA Engine
Inspect the product boundary →Showcase · evidence surface
Show me what you actually built.
Showcase is cross-category. Each entry identifies what you are seeing, what produced it, what kind of evidence it is, and where to inspect more. A concept image is never silently presented as implementation evidence.
Creation
Authoring and runtime surfaces.
Creation evidence shows the relationship between the tool, the runtime, and the artifact. It does not require every project to share an owner or distribution model.
Cartographer spatial authoring
Revision-bound geometry and topology authoring with validation and neutral interchange contracts.
Produced byCartographer
Inspect the authoring boundary →Worlds
A stack with visible handoffs.
CONDITION is a case study rather than a reason to collapse Cartographer and VANTA into one product.
CONDITION creation stack
Cartographer authors space. VANTA runs the world. CONDITION is the experience built on top.
Evidence boundaryCase study; implementation status varies by feature.
Read the relationship →OMNI-Q simulated development
Bounded routing, verification, recovery, and evidence receipts in a simulation-oriented development surface.
Not claimedHardware performance or universal trained-model capability.
Inspect the source boundary →AI & training
Receipts stay separate from promises.
Model and training entries identify their evidence tier. A release page, benchmark, or native execution path does not become a promotion or deployment claim by appearing here.
Neural Foundry execution boundary
Local execution, checkpoint, integrity, and bounded Hub coordination described without exposing private runtime details.
SourceCanopy-Foundry public boundary
Read the boundary →HAVEN local-first surface
Household-scoped history, provider seams, speech, local model management, and authority-gated action paths.
SourceHAVEN public repository
Read the product page →TraceGlass decision-chain reconstruction
A diagnostic lens for reconstructing information, belief, action, and consequence.
StatusResearch overview, not a universal log viewer claim.
Read the research page →Evaluation
Test the behavior, not the branding.
The evaluation surfaces keep host truth, model output, policy, replay, and outcome evidence separate.
Open World Model Harness
Long-horizon evaluation under partial information with evaluator-held ground truth and host-owned world state.
Evidence typeEvaluation harness
Read the project overview →Ghost Teacher capability probing
An adaptive evaluator and curriculum engine direction for locating failure boundaries without calling itself a trainer.
Evidence typeResearch direction
Read the research overview →Truth discipline
If it is not measured, it is not a measurement.
Each card on this page names the project, evidence type, implementation status, and a detail path. Follow the source receipt before extending the claim.