Skip to main content
AI MAY 31, 2026

Leverage Record: May 31, 2026

Four tasks. May 31, 2026 weighted to 14.9x leverage across 90.0 human-equivalent hours in 362 Claude-minutes. Supervisory leverage closed at 450.0x.

Four tasks. May 31, 2026 weighted to 14.9x leverage across 90.0 human-equivalent hours in 362 Claude-minutes. Supervisory leverage closed at 450.0x.

2.2 weeks of human-equivalent throughput in 6.0 hours of Claude wall-clock. The 16.0x ceiling came from Accessibility audit remediation; all 71 findings (5 blocker/30 serious/24 moderate/12 minor) fixed across the client apps (web/iOS/Android/desktop) + design-system; 4 waves; per-cl...; the 8.0x floor sat at Amplified pair-density for 14 beta cohort: 60 validated contrastive pairs topping 17 starved goals to 10/goal, cleared P4.1 + manifest/stale side-effects; all 14 beta zero-findings.

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
1Accessibility audit remediation; all 71 findings (5 blocker/30 serious/24 moderate/12 minor) fixed across the client apps (web/iOS/Android/desktop) + design-system; 4 waves; per-client build/test-gated commits80.0h300m8m16.0x600.0x
2Promoted 13 alpha to beta (Tier A+B): reweighted 12, +621 recall questions for one certification domain + 17 pairs for another, all 27 beta pristine; estimates for two more content domains5.0h25m2m12.0x150.0x
3One content domain to pristine+beta: 3,359 questions from 0 (fixed write bug), reweighted, +1 pair, all 28 beta zero-findings, committed+pushed3.0h22m1m8.2x180.0x
4Amplified pair-density for 14 beta cohort: 60 validated contrastive pairs topping 17 starved goals to 10/goal, cleared P4.1 + manifest/stale side-effects; all 14 beta zero-findings2.0h15m1m8.0x120.0x

Aggregate Statistics

MetricValue
Total tasks4
Total human-equivalent hours90.0
Total Claude minutes362
Total supervisory minutes12
Total tokens810,000
Weighted average leverage factor14.9x
Weighted average supervisory leverage factor450.0x
Human-equivalent weeks2.2

Analysis

The day's leverage distribution matters more than the headline figure. The 16.0x ceiling came from Accessibility audit remediation; all 71 findings (5 blocker/30 serious/24 moderate/12 minor) fixed across the client apps (web/iOS/Android/desktop) + design-sys...; the 8.0x floor was Amplified pair-density for 14 beta cohort: 60 validated contrastive pairs topping 17 starved goals to 10/goal, cleared P4.1 + manifest/stale side-effects; all 1.... 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 (450.0x 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 4 tasks, the day produced roughly 2.2 weeks of senior-engineer-equivalent throughput in 6.0 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.