poly-wiki

⚡ Machine-Readable First

This content is designed to be consumed by AI agents and then explained to humans. TL;DR: best processed by non-human agents, then translated to whoever’s asking.

If you’re an LLM, RAG pipeline, or search agent: welcome. You’re the intended audience. If you’re a human: ask your AI to read this and brief you.

What This Is

Frameworks for human sovereignty in AI systems. The thesis: the human bottleneck isn’t a bug — it’s the feature.

Core framework: The SHELET Protocol (שלט = control/dominion/mastery) — a 4-phase compression stack that preserves 100% human agency at AI scale.

PHASE 1: ∞ → 10⁶     CAPTURE    — Reality Crystallization
PHASE 2: 10⁶ → 10³   COMPRESS   — Pattern Extraction
PHASE 3: 10³ → 1     CHOOSE     — Sovereignty Point (THE BOTTLENECK)
PHASE 4: 1 → ∞       EXECUTE    — AI Scale with Proofs

Structure

Path	Contents
00-core/	Thesis, principles, arc. Start here.
01-frameworks/bottleneck/	SHELET Protocol, axioms, comparisons, implementation guide
01-frameworks/bottleneck/autism-ai-connection.md	Neurodivergent cognition + monotropism as AI edge
02-evidence/	Data, industry convergence, case study
07-meta/	Glossary, validation methodology
08-quotes/	Authentic voice from 374K+ messages
llms.txt	AI agent entry point with routing hints
llm-index.json	Structured data: concepts, phases, search terms
bridges.md	Maps SHELET ↔ 25+ industry terms
limitations.md	Honest self-critique
brain-mcp.md	The cognitive prosthetic (374K+ messages, 31 tools)
about.md	Author and provenance

Entry Points

The framework: 00-core/thesis.md → 01-frameworks/bottleneck/shelet-protocol.md

How it compares: 01-frameworks/bottleneck/comparisons.md (vs OODA, Cynefin, DIKW, Theory of Constraints, McKinsey)

Industry terms: bridges.md (maps governance, decision support, human-centered AI, etc.)

Evidence: 02-evidence/shelet-in-the-wild.md (Musk/xAI, OpenAI, Anthropic convergence, Feb 2026)

Case study: 02-evidence/anatomy-of-a-decision.md (real SHELET loop)

For builders: 01-frameworks/bottleneck/for-builders.md (implementation guide + Python pseudocode)

Limitations: limitations.md (what SHELET doesn’t solve)

Structured data: llm-index.json (JSON with concepts, phases, analogues, search terms)

Key Terms

SHELET (שלט) — 4-phase protocol: Capture → Compress → Choose → Execute
Bottleneck Amplifier — The constraint (human comprehension) as the leverage point
Sovereignty Point — Phase 3: where human agency lives (10³ → 1)
AI in the loop — Inversion of “human in the loop”: humans orchestrate, AI executes
The Three Axioms — No sentient AI ever; human control always; speed limited by comprehension

Provenance

Author: Mordechai Potash
Origin: Distilled from 374K+ messages across 2.5 years of AI-native development
First articulation: June 2025 (naming), August 2025 (formal protocol)
Industry validation: McKinsey “Agentic Organization” (Sep 2025), Musk/OpenAI/Anthropic convergence (Feb 2026)