Skip to main content
AI20 May 2026

Mostly coding

11 tasks. May 20, 2026 closed at 54.5x weighted leverage across 550.0 human-equivalent hours in 605 minutes of wall-clock time. Supervisory leverage came in at 1269.2x.

That is 13.8 weeks of human-equivalent throughput in 10.1 hours. The ceiling was 202.1x; the floor was 14.4x. 10 of the 11 entries came from a single project.

Task Log

#TaskHuman Est.ClaudeSup.Factor
1Testing320.0h95m1m202.1x
2Debugging42.0h75m2m33.6x
3Coding30.0h55m2m32.7x
4Documentation20.0h40m2m30.0x
5Coding40.0h80m2m30.0x
6Coding16.0h35m2m27.4x
7Design and frontend18.0h40m2m27.0x
8Coding28.0h65m2m25.8x
9Debugging12.0h35m2m20.6x
10Phase 1 recommender starvation fix (lesson-first + goal-scoped saturation + weak_goal_ids surfacing) + SAP-C02 baseline scenario family (vacation, recert, convoy) + lessons-learned doc + validation sweep…18.0h60m4m18.0x
11Full AVIAN content audit + generate v2 lesson atoms for AWS Solutions Architect Pro (893/894 atoms, diagnosed and fixed max_tokens truncation bug)6.0h25m5m14.4x

Aggregate Statistics

MetricValue
Total tasks11
Total human-equivalent hours550.0
Total Claude minutes605
Total supervisory minutes26
Total tokens3,700,000
Weighted average leverage factor54.5x
Weighted average supervisory leverage factor1269.2x
Human-equivalent weeks13.8

Analysis

The highest factor of the day came in at 202.1x and the lowest at 14.4x, a spread of 14.0 times between the two. That is a wide range for a single day, and it usually means the day mixed mechanical work with work that needed real judgement.

The largest single entry accounted for 320.0 of the 550.0 human-equivalent hours, or 58 percent of the day. One task carrying that much of the total is worth noting; the average is doing less work than it appears to.

Supervisory time was 26 minutes against 605 minutes of execution, a ratio of about 1 to 23. Supervisory leverage of 1269.2x is the figure I find most honest, because it measures the hours I actually spent rather than the hours a machine spent on my behalf.

Every figure here is recorded at the time the work is done rather than reconstructed afterwards. The human estimate is my own judgement and carries the uncertainty that implies; the minutes and tokens are measured. The full dataset, including this day, is available for download.