01 v1.7
Context Is the Universal Bottleneck Context architecture determines reasoning trajectories, not just outputs — context architecture IS attention architecture across both silicon and carbon substrates.
8 mechanisms · 7 evidence chains Explore →
02 v1.7
Human Judgment Remains the Integration Layer AI handles tasks, humans integrate into value. Evaluation migrates upward from discriminative taste to situated governance — purpose, history, consequence, quality. The judgment layer grows MORE important as generation compresses, but the population capable of exercising it is shrinking.
7 mechanisms · 6 evidence chains Explore →
03 v1.6
Architecture Matters More Than Model Selection Format > harness > prompt > model. The surrounding system dominates the model inside it. Models trained agentic-first REQUIRE rich context architecture — baked into weights, creating co-evolutionary self-reinforcement.
7 mechanisms · 5 evidence chains Explore →
04 v1.3
Build Infrastructure to Delete Invest in durable boundaries around volatile capabilities. Expect 80% deletion — make it cheap.
4 mechanisms · 3 evidence chains Explore →
05 v1.9
Orchestration Is the New Core Skill Management skills transfer to AI orchestration. The management-implementation boundary is dissolving from BOTH directions — developers becoming PMs and PMs becoming developers. Coordination overhead absorbs 60-70% of individual productivity gains.
7 mechanisms · 5 evidence chains Explore →
06 v3.0
Speed and Knowledge Are Orthogonal Zero friction can mean zero knowledge. The Evaluation Crisis Spiral: generation speed → evaluation debt → fatigue → verification decline → quality floor collapse → disposability → more speed. Validated across 9 domains including education, peer review, and manufacturing.
8 mechanisms · 6 evidence chains Explore →
