Skip to main content
AI MAY 28, 2026

Leverage Record: May 28, 2026

Three tasks. May 28, 2026 weighted to 49.9x leverage across 104.0 human-equivalent hours in 125 Claude-minutes. Supervisory leverage closed at 416.0x.

Three tasks. May 28, 2026 weighted to 49.9x leverage across 104.0 human-equivalent hours in 125 Claude-minutes. Supervisory leverage closed at 416.0x.

2.6 weeks of human-equivalent throughput in 2.1 hours of Claude wall-clock. The 168.0x ceiling came from Built comprehensive content-domain curriculum: 4 domain specs (Math/RLA/Science/Social Studies) totaling 249 leaf goals plus README with 16 novel adult-learner activities.; the 16.0x floor sat at Resume Phase 0 of a security-scanning service; verify backend pytest plus frontend build plus git init/commit plus reserve ports; then a 48-agent adversarially-verified project rev....

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
1Built comprehensive content-domain curriculum: 4 domain specs (Math/RLA/Science/Social Studies) totaling 249 leaf goals plus README with 16 novel adult-learner activities.70.0h25m3m168.0x1400.0x
2Resume session: re-stamped 584 spec/manifest exam_metadata fields across 142 packages plus restamp tool plus 3 recall backfills (+2555 questions) plus content-audit Phase 27 plus P25.7 plus audit rerun (144 findings) plus 2-agent synthesis/engine field-coverage investigation.22.0h55m8m24.0x165.0x
3Resume Phase 0 of a security-scanning service; verify backend pytest plus frontend build plus git init/commit plus reserve ports; then a 48-agent adversarially-verified project review and fix of every confirmed broken-now/Phase-0 finding (TS2559 dead LoginPage; phantom defect-reporter dep; CSS @import order; vitest passWithNoTests; /me 401; AUTH_DISABLED default; pinned Trivy; ModalProvider; FOUC else-branch; branded favicons; container-registry docker token; backend tests to 98% coverage; docs); verified green and committed locally.12.0h45m4m16.0x180.0x

Aggregate Statistics

MetricValue
Total tasks3
Total human-equivalent hours104.0
Total Claude minutes125
Total supervisory minutes15
Total tokens3,180,000
Weighted average leverage factor49.9x
Weighted average supervisory leverage factor416.0x
Human-equivalent weeks2.6

Analysis

The day's leverage distribution matters more than the headline figure. The 168.0x ceiling came from Built comprehensive content-domain curriculum: 4 domain specs (Math/RLA/Science/Social Studies) totaling 249 leaf goals plus README with 16 novel adult-learner...; the 16.0x floor was Resume Phase 0 of a security-scanning service; verify backend pytest plus frontend build plus git init/commit plus reserve ports; then a 48-agent adversarially-.... 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 (416.0x 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 3 tasks, the day produced roughly 2.6 weeks of senior-engineer-equivalent throughput in 2.1 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.