Skip to main content
AI MAY 18, 2026

Leverage Record: May 18, 2026

Five tasks. May 18, 2026 weighted to 30.4x leverage across 190.0 human-equivalent hours in 375 Claude-minutes. Supervisory leverage closed at 518.2x.

Five tasks. May 18, 2026 weighted to 30.4x leverage across 190.0 human-equivalent hours in 375 Claude-minutes. Supervisory leverage closed at 518.2x.

4.8 weeks of human-equivalent throughput in 6.2 hours of Claude wall-clock. The 120.0x ceiling came from Review an admin client and author full Stitch prompt for Westworld Delos-themed WebGL/Rive redesign covering all 24 pages, design tokens, component vocabulary, motion language, aud...; the 13.6x floor sat at Docstring audit Phase 7 (Protocol contract enforcement): new audit script (scripts/auditprotocolcontracts.py, 857 LoC) with AST-based one-hop expansion through same-class helpers....

About These Records
These time records capture personal project work done with Claude Code (Anthropic) only. They do not include work done with ChatGPT (OpenAI), Gemini (Google), Grok (xAI), or other models, all of which I use extensively. Client work is also excluded, despite being primarily Claude Code. The actual total AI-assisted output for any given day is substantially higher than what appears here.

Task Log

#TaskHuman Est.ClaudeSup.FactorSup. Factor
1Review an admin client and author full Stitch prompt for Westworld Delos-themed WebGL/Rive redesign covering all 24 pages, design tokens, component vocabulary, motion language, audio design, and fidelity grading rubric24.0h12m3m120.0x480.0x
2Aperture V2 viewer Phases 1-3: Three.js stage layer (paper-grain + page-turn shaders, mastery candle, postprocessing), Rive Living Diagrams integration (validator update in engine), layer-registry slot system, Settings UI toggle, 19 vitest cases, Storybook themes/density stories45.0h35m1m77.1x2700.0x
3Resume + commit cleanup across engine/audits/domains, then scaffold Phase 0 Aperture V2 lesson viewer (4-layer V1↔V2 toggle, theme + motion + density registries, ApertureShell, AdaptiveDensityLayer idea #9, ViewerErrorBoundary, an analytics platform telemetry; ~950 LOC; vite build clean)16.0h28m6m34.3x160.0x
4an inference engine autopilot Fix A: coverage damping + hard ceiling on readiness. SOA-C02 baseline 36/73 KG goals at exampassed→ 73/73 covered + passed; 36/38 cloud certs hit full per-goal coverage across AWS/GCP/Azure cascade. computedomainreadiness (helpers.py:712+) and computenext_actions (autopilot...100.0h278m10m21.6x600.0x
5Docstring audit Phase 7 (Protocol contract enforcement): new audit script (scripts/auditprotocolcontracts.py, 857 LoC) with AST-based one-hop expansion through same-class helpers AND field-attribute delegates; audited 12 raise contracts across 7 Protocol abc.py files against 7 canonical implementer classes;...5.0h22m2m13.6x150.0x

Aggregate Statistics

MetricValue
Total tasks5
Total human-equivalent hours190.0
Total Claude minutes375
Total supervisory minutes22
Total tokens1,416,500
Weighted average leverage factor30.4x
Weighted average supervisory leverage factor518.2x
Human-equivalent weeks4.8

Analysis

The day's leverage distribution matters more than the headline figure. The 120.0x ceiling came from Review an admin client and author full Stitch prompt for Westworld Delos-themed WebGL/Rive redesign covering all 24 pages, design tokens, component vocabulary,...; the 13.6x floor was Docstring audit Phase 7 (Protocol contract enforcement): new audit script (scripts/auditprotocolcontracts.py, 857 LoC) with AST-based one-hop expansion throug.... Tasks at the top of the distribution share a shape: tightly-scoped specifications, clear success criteria, and minimal integration ambiguity. The AI doesn't need to discover anything new; it executes against an explicit target.

Tasks at the bottom run differently. They're either bounded by review-heavy work where every step gets verified, or they involve ambiguity that demands several rounds of trial and adjustment. The factor is real and informative, not a failure mode.

The supervisory leverage figure (518.2x today) tracks something orthogonal to wall-clock leverage. It's the ratio of human-equivalent output to human prompt-writing time. It stays high even on lower-leverage days because supervisory minutes scale with task count, not with the human-hour estimate; a 20-minute task and a 4-hour task can both be specified in two minutes of human prompt-writing.

Across the 5 tasks, the day produced roughly 4.8 weeks of senior-engineer-equivalent throughput in 6.2 hours of model wall-clock. That ratio is the practical answer to the question of how much output a single operator can move per day when the model handles the execution and the operator handles the direction.