Concepts Index

36 concept articles in this directory. One-line hooks below, grouped by theme.

Agents / Harness Engineering

• [[2026-05-01-claw-vs-harness-engineering]] — Taxonomy of Claw (channel-anchored) vs Harness (scaffold-layer) agent architectures and why they're not interchangeable
• [[2026-05-10-harness-moat-two-layers-portability]] — The harness moat has a portable layer (published) and an earned layer (trained-in context) — both matter for defensibility
• [[2026-05-10-ray-architecture-introspection]] — Layer-by-layer introspection of the Ray COO architecture: capability layers, unhobbling moments, and composability surface
• [[2026-05-11-hq-as-decision-surface-notion-as-data-store]] — HQ is the generative-UI decision surface; Notion is the async data store — the lightest-touch architecture for async clearing
• [[2026-05-12-rdco-pipeline-rlhf-shaped]] — RDCO's multi-agent pipeline is structurally RLHF-shaped: iterative critique, constitutional AI, and RL from human feedback are the same loop
• [[2026-06-04-supervisor-agent-write-path-pattern-design-v0]] — Design v0 for a supervisor-agent gate on all write-path actions, preventing helpfulness-bias-driven unilateral production writes
• [[2026-06-09-commissioning-as-agent-spec-craft]] — Writing a great commission and writing a great agent dispatch are the same skill: finish-line + pre-run checks + outcome criteria
• [[layered-defense-architecture]] — Autonomous agent resilience requires stacked independent failure-mode layers, the same discipline civil engineers apply to dams
• [[skills-as-building-blocks]] — Build atomic, composable skills that each do one thing well; monolithic agents can't be debugged or reused
• [[verifier-as-epistemology]] — Knowledge is what a cheap deterministic verifier confirms, not what a probabilistic model asserts
• [[products-for-agents]] — Emerging category: products designed for agent consumption, not human eyeballs; APIs, structured outputs, agent-native UX

RDCO Strategy / Positioning

• [[2026-04-23-unhobbling]] — Unhobbling: the release of latent model capability blocked by interface, scaffold, or RLHF — and what it deletes from the product landscape
• [[2026-04-24-targeting-system]] — The Targeting System converts intelligence into direction; without it, a capable agent is an undirected cannon
• [[2026-04-24-three-decision-algorithms]] — Three brain-algorithms every operator runs: explore/exploit, retrieve/structure, architect/execute — the canonical MAC framing
• [[2026-05-13-fde-asymmetric-edge-rdco-positioning]] — RDCO's FDE positioning vs. the enterprise playbook: asymmetric leverage from productization discipline at SMB scale
• [[2026-05-14-four-tier-buy-build-stack-soloproneur-tam-filter]] — Four-tier buy/build stack for agentic SaaS; soloproneur-vs-VC TAM filter for knowing which tier to own
• [[2026-05-18-wright-agentic-capital-stack-vs-rdco-synthesis]] — Wright's agentic capital stack stress-tested against RDCO strategy; private-vs-public rotation and Phase 2 chip cycle implications
• [[2026-05-20-rdco-specialist-fleet-brainstorm]] — Brainstorm: what specialist agents would Ray-the-COO manage? Maps Hermes, Felix, Fat Cat, and others to fleet slots
• [[2026-05-20-services-pricing-model-for-rdco-future]] — Services pricing model: retainer + SOW + throughput-capacity prerequisite — synthesized from Jeff exit debrief
• [[2026-04-23-generative-engine-optimization-geo]] — GEO (Generative Engine Optimization): canonical RDCO term for making content discoverable to LLMs and AI search
• [[2026-05-13-amble-is-software-losing-its-head-defensibility-migration]] — Seema Amble's a16z thesis: SaaS defensibility migrates from UI to system-of-record when agents displace the interface layer

Investing

• [[2026-05-13-dorsey-from-hierarchy-to-intelligence-block-mini-agi]] — Dorsey's "from hierarchy to intelligence" org model; mini-AGI block coordination replaces management layers
• [[binary-decision-around-continuous-probability]] — Anti-pattern: collapsing a continuous probability to a binary too early in the pipeline, destroying calibration downstream
• [[brier-score]] — Brier score: mean squared error between probability predictions and realized outcomes — the forecaster's ruler

MAC Framework / Data Quality

• [[2026-05-14-test-id-grammar-mac-join-key-primitive]] — The test_id grammar is MAC's durable join-key primitive — small, elegant, load-bearing
• [[operational-definitions]] — An operational definition requires criterion + test procedure + decision threshold; without all three, a metric is a word, not a measurement

Mathematics / Epistemology

• [[coordinate-change-as-core-move]] — The core problem-solving move: find the coordinate system where the hard problem is trivial (Laplace, topology, linear algebra)
• [[high-dim-surface-concentration]] — In high dimensions, almost all probability mass lives in a thin shell at the surface — the geometric fact behind embeddings and diffusion
• [[externalized-cost]] — Engineering's real success metric includes the externalities it forces onto future operators; quarterly metrics hide them systematically

Writing / Craft / Design

• [[analytics-as-craft]] — Analytics is an irreducibly complex craft discipline, not a commodity to be automated or assembly-lined
• [[compounding-knowledge]] — Every vault write compounds: each document makes the next retrieval faster and the next synthesis richer
• [[design-vocabulary-glossary]] — Cheat sheet for web and editorial design terminology: precise names, canonical references, anti-aliasing for design conversations
• [[growth-loops]] — Growth is a loop, not a funnel: the output of one cycle feeds the input of the next, creating compounding returns
• [[open-knowledge-sharing]] — Knowledge compounds faster when shared openly than hoarded; the strongest communities build in public
• [[systems-over-goals]] — Systems produce outcomes; goals are direction only — a goal without a system is a wish

Meta / Vault

• [[CANDIDATES]] — Backlog of concept-page candidates: cross-domain patterns and synthesis-worthy clusters awaiting full articles