AgDR-Phoenix v1.8 · April 2026
The Zero-Latency Floor.
The Atomic Genesis Decision Record (AgDR) is the open standard for court-admissible AI decision records. Sub-microsecond provenance tracking for forensic AI integrity.
QPC: ML-DSA
BONTLI_COMPRESSION
LATENCY
0.62 µs
THROUGHPUT
1.1M/s
INTEGRITY
BLAKE3
RELIABILITY
100%
Statutory Teeth
CEA s.31.1 AdmissibilityCanada Evidence Act
CBCA s.122 MappingDirector Fiduciary Duty
Trust the Record.
The AI accountability kernel is finding its audience. 955 clones in 11 days with zero marketing.
Provenance: Who is acting, on whose behalf.
Place: Where the decision is headed.
Purpose: The explicit intent and ethical anchor.
Place: Where the decision is headed.
Purpose: The explicit intent and ethical anchor.
{
"version": "1.8.2",
"aki_latency_ns": 620,
"sig": "ml-dsa_qpc...",
"root": "blake3_forest..."
}
What is an AI decision record?
It is a permanent, tamper-proof log captured at the exact inference instant. Think of it as a flight data recorder for AI reasoning. Every decision is timestamped, cryptographically signed, and immutable.
Does this slow down innovation?
No. At 620 nanoseconds, it is faster than a network packet. It removes the governance friction that keeps institutional capital on the sidelines.
What is the AgDR standard?
AgDR (Atomic Genesis Decision Record) is an open standard for creating court-admissible AI decision records. It provides cryptographic proof of AI reasoning with sub-microsecond latency, compliant with EU AI Act, NIST AI RMF, and Canada Evidence Act.
How does AgDR ensure data integrity?
AgDR uses BLAKE3 hashing and ML-DSA quantum-resistant signatures to create tamper-evident records. The zero-latency design means records are locked in immediately at inference time, preventing retroactive modification.
Is AgDR suitable for regulated industries?
Yes. AgDR is designed specifically for sectors requiring forensic auditability—financial services, healthcare, legal, and government. Its compliance with the Canada Evidence Act and EU AI Act makes it court-admissible.
How do I get started?
Visit our Getting Started guide or try the Phoenix Demo sandbox. We provide libraries for Rust (crates.io), Python (PyPI), and integration with existing ML frameworks. Full documentation is available on GitHub.