Draft Living Standard

AI-Native Architecture

A family of specifications defining what it means for a software system to treat AI agents as first-class architectural participants — with identity, permissions, memory, observability, and agency — not bolted-on features.

Reverse-derived by design. Every normative requirement traces to a concrete anti-pattern observed in production. No requirement exists without a corresponding failure mode.

AI-Native Architecture
CORE · 15 §

The foundation. Defines AI-native systems that grant agents the same architectural standing as human users: shared memory, identity, calibration and closed loops, context propagation, safety gates, policy at every touchpoint, runtime interoperability, observability, and an anti-pattern catalog.
Read specification →
Agent Construction & Coordination
AGENTS

How AI agents are constructed, deployed, evaluated, and coordinated on the AI-native foundation. Covers what an agent is, how it reasons and acts, and how multiple agents operating on the same system coordinate safely.
Read specification →
AI-Native Self-Evolution
EVOLUTION

How an AI-native system autonomously improves itself. With identity, memory, calibration, safety gates, and observable agents already in place, self-evolution closes the loop — the system proposes, evaluates, and adopts changes to its own behavior under guardrails.
Read specification →