01 / SIGNAL TRUST
Should this observation enter computation?
Evidence about the source must exist before downstream compute can rely on it.
Living Cipher Lab · Public Demonstrators
Working evidence, not architecture theater.
Living Cipher Lab is where architectural claims are turned into observable systems. Each demonstrator addresses a different boundary in the compute trust chain: whether a signal, computational participant, or execution environment has earned admission — and whether evidence of that decision can survive into the result.
Source → Trust → Compute → Hardware → Attestation
Three Trust Boundaries
The question is not simply whether something can execute.
The question is what deserves admission into a consequential computing system.
02 / COMPUTE TRUST
Should this component participate?
Heterogeneous compute participants require identity, policy, and admission — not implicit trust by presence.
03 / EXECUTION TRUST
Can execution be bound to authorized hardware?
Hardware identity, approved state, freshness, and execution evidence must survive into the receipt.
Demo 01 · Live-Source Trust · Public Prototype
Before computation can be trusted, the signal has to earn admission.
The current public demonstrator captures a live voice-channel signal locally in the browser, derives a multidimensional nonlinear witness state from observed dynamics, and binds that evidence to the capture session to produce an attestation receipt. A subsequent observation produces its own evidence state and receipt, allowing changed source conditions to become visible at the receipt level.
Live Signal
Witness State
Admission Evidence
Attestation Binding
Receipt
Public boundary
The interface deliberately abstracts the underlying witness methods. The prototype does not claim to detect every possible manipulation; it demonstrates that changes in observed source conditions can change the evidence used to admit and attest the observation.
Evidence Convergence
Evidence should not depend on a single view of the source.
A primary witness can characterize properties of the observed signal itself. Living Cipher's broader architecture also supports ancillary evidence — independent or auxiliary observations that test whether a signal remains consistent with the process, device, environment, or physical state expected to have generated it.
The purpose is not to accumulate arbitrary metrics. It is to establish independent evidence convergence before admission.
Demo 02 · Compute Admission · Public Simulation
A connected component has not yet earned the right to participate.
The Chiplet Trust simulation visualizes component enrollment, verification, refusal, quarantine, isolation, and attribution across a heterogeneous multi-vendor compute package. The public surface shows protocol behavior while keeping keys, enrollment criteria, deviation logic, and hardware-root implementation details private.
Component
Identity
Policy
Admission
Execution
Connectivity is not admission.
Physical or logical connectivity does not confer participation rights. Identity, approved state, policy compatibility, and authenticated verdicts remain explicit architectural controls.
Architecture Trajectory
Trust does not stop at the signal boundary.
Modern systems distribute computation across CPUs, GPUs, NPUs, DSPs, programmable fabric, accelerators, chiplets, remote infrastructure, and specialized silicon. Each participant creates another admission decision.
CURRENT PUBLIC DEMO
Live-Source Trust
Evidence about the observed source before downstream computation relies on it.
Source → Witness → Evidence → Receipt
PUBLIC SIMULATION
Compute Admission
Identity, authorized state, policy, quarantine, and participation boundaries for heterogeneous compute.
Component → Identity → Policy → Admission
HARDWARE INTEGRATION PATH
Hardware-Rooted Trust
A fresh challenge bound to device identity, authorized state, execution, and a hardware-rooted receipt.
Challenge → Hardware → Execution → Receipt
The Architecture Converges
The objective is a single evidence chain.
Living Cipher is building toward a system in which trust can be carried from the condition of the source, through the components allowed to compute on it, into the hardware that executes the computation, and finally into the provenance of the result.
Biological Source
Signal Admission
Compute Admission
Authorized Hardware
Result Provenance
Nothing earns trust merely by being present.
Signals earn admission. Compute components earn admission. Hardware establishes execution evidence. Functions earn their place in silicon.