42 tasks. July 1, 2026 closed at 29.4x weighted leverage across 518.5 human-equivalent hours in 1,059 minutes of wall-clock time. Supervisory leverage came in at 198.2x.
That is 13.0 weeks of human-equivalent throughput in 17.6 hours. The ceiling was 80.0x; the floor was 5.0x. 40 of the 42 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | Author Civics and Government knowledge taxonomy for AccelaStudy AI | 8.0h | 6m | 3m | 80.0x |
| 2 | Author 94 alpha domain specs (29 languages, 10 AP gaps, 22 IB gaps, 30 K-12 grade courses, 3 pilots) via web-grounded 91-agent fan-out + generalized/hardened author_spec.py assembler; all pass structural… | 120.0h | 95m | 5m | 75.8x |
| 3 | Author 115 vertical certification domain specs via 2-wave web-grounded fan-out + vendor-canonical dedup gate; survived mid-run quota interrupt with zero content loss (external backup + git); all pass… | 150.0h | 120m | 5m | 75.0x |
| 4 | The Deferral: 3-round multi-agent editorial review (30 full-manuscript reads) + surgical revision (fair-play plant, thread closures, HERA seeding, opening rework, authenticity patches, 2 prose tic passes) +… | 70.0h | 90m | 4m | 46.7x |
| 5 | Greek 101 knowledge taxonomy for AccelaStudy AI Languages (CEFR A1, 68 leaf goals, 8 domains) | 6.0h | 12m | 4m | 30.0x |
| 6 | CAIA Level II knowledge taxonomy authorship | 6.0h | 12m | 3m | 30.0x |
| 7 | Author EX280 Red Hat OpenShift Specialist knowledge taxonomy with web research | 8.0h | 18m | 5m | 26.7x |
| 8 | The Deferral: full 15-reader editorial review + publication-prep structural revision (Ch31 split, boardroom scene, HERA interstitial, Finn cost arc, tic pass, Book-2 seeds, audit+docs) | 40.0h | 105m | 3m | 22.9x |
| 9 | Author USMLE Step 2 CK knowledge taxonomy for AccelaStudy AI | 3.0h | 8m | 3m | 22.5x |
| 10 | Czech 101 knowledge taxonomy 68 leaf goals CEFR A1 full prereq graph | 4.0h | 12m | 4m | 20.0x |
| 11 | Author Hungarian 101 knowledge taxonomy (CEFR A1) — 64 leaf goals, 89 prereq edges, 6 domains | 4.0h | 12m | 3m | 20.0x |
| 12 | Author CFP exam knowledge taxonomy (70 leaf goals, 8 domains) | 4.0h | 12m | 3m | 20.0x |
| 13 | Author QSBA Qlik Sense Business Analyst knowledge taxonomy - 72 leaf goals across 4 official exam domains | 4.0h | 14m | 5m | 17.1x |
| 14 | Author HashiCorp Vault Associate 002 exam knowledge taxonomy (63 leaf goals, 8 domains) | 6.0h | 22m | 5m | 16.4x |
| 15 | Author Swedish 101 CEFR A1 knowledge taxonomy for AccelaStudy AI Languages — 8 domains, 72 leaf goals | 3.0h | 12m | 3m | 15.0x |
| 16 | Grade 6 Mathematics CCSS knowledge taxonomy for AccelaStudy AI | 3.0h | 12m | 3m | 15.0x |
| 17 | Author OSEP PEN-300 knowledge taxonomy for AccelaStudy AI (80 nodes, 69 leaf goals, 14 prereq edges) | 3.0h | 12m | 3m | 15.0x |
| 18 | Author Turkish 101 knowledge taxonomy for AccelaStudy AI Languages (68 leaf goals, CEFR A1, ACTFL Novice) | 4.0h | 18m | 5m | 13.3x |
| 19 | Author IB Theatre HL knowledge taxonomy with web research | 4.0h | 18m | 5m | 13.3x |
| 20 | IB World Religions HL taxonomy authoring - 70 leaf nodes 9 domains | 4.0h | 18m | 3m | 13.3x |
| 21 | Author RHCA knowledge taxonomy for AccelaStudy AI | 4.0h | 18m | 5m | 13.3x |
| 22 | CIA exam knowledge taxonomy authoring (3-part IIA CIA blueprint research + 70-node flat node list) | 4.0h | 18m | 3m | 13.3x |
| 23 | Thin avian-engine to code+docs: hard-gated removal of 49G package content+caches (containment analysis vs canonical+S3, 683M external backup, per-item divergence diff) + root-cause fix of relative-path… | 4.0h | 20m | 3m | 12.0x |
| 24 | Author IB English Literature HL knowledge taxonomy — 73 nodes, 8 domains, 12+ prereq edges, web research | 3.5h | 18m | 3m | 11.7x |
| 25 | Author CIPP/US knowledge taxonomy for AccelaStudy AI (60-80 leaf goals across 5 exam domains) | 4.0h | 22m | 3m | 10.9x |
| 26 | Hebrew 101 knowledge taxonomy for AccelaStudy AI Languages | 3.0h | 18m | 3m | 10.0x |
| 27 | Author Finnish 101 CEFR A1 knowledge taxonomy for AccelaStudy AI Languages — 8 domains 72 leaf goals | 3.0h | 18m | 3m | 10.0x |
| 28 | Ukrainian 101 knowledge taxonomy authorship | 3.0h | 18m | 5m | 10.0x |
| 29 | AP Research knowledge taxonomy authorship | 3.0h | 18m | 3m | 10.0x |
| 30 | Author Databricks DE-P knowledge taxonomy with web research and official exam guide extraction | 3.0h | 18m | 3m | 10.0x |
| 31 | Author GPYC GIAC Python Coder knowledge taxonomy with web-researched exam blueprint | 2.0h | 12m | 3m | 10.0x |
| 32 | Author CAPA knowledge taxonomy for AccelaStudy AI | 3.0h | 18m | 5m | 10.0x |
| 33 | SHRM-SCP knowledge taxonomy authoring with web research | 3.0h | 18m | 5m | 10.0x |
| 34 | Author Praxis Core 5712-5722-5732 knowledge taxonomy for AccelaStudy AI | 2.5h | 18m | 5m | 8.3x |
| 35 | AP Seminar knowledge taxonomy (68 leaf goals QUEST framework) | 3.0h | 22m | 3m | 8.2x |
| 36 | Author CPIM knowledge taxonomy for AccelaStudy AI | 3.0h | 22m | 5m | 8.2x |
| 37 | Author Alteryx Designer Core (ADC) knowledge taxonomy - 69 leaf goals across 4 exam domains | 2.0h | 18m | 5m | 6.7x |
| 38 | Author CIPM knowledge taxonomy for AccelaStudy AI | 2.0h | 18m | 3m | 6.7x |
| 39 | CCSK taxonomy authoring - CCSK v5 12-domain knowledge map with 70 leaf goals | 2.0h | 18m | 3m | 6.7x |
| 40 | Author CFA Level III knowledge taxonomy for AccelaStudy AI | 2.0h | 18m | 3m | 6.7x |
| 41 | Coding | 4.0h | 45m | 3m | 5.3x |
| 42 | Author CIS-HRSD knowledge taxonomy for AccelaStudy AI | 1.5h | 18m | 3m | 5.0x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 42 |
| Total human-equivalent hours | 518.5 |
| Total Claude minutes | 1,059 |
| Total supervisory minutes | 157 |
| Total tokens | 27,965,000 |
| Weighted average leverage factor | 29.4x |
| Weighted average supervisory leverage factor | 198.2x |
| Human-equivalent weeks | 13.0 |
Analysis
The highest factor of the day came in at 80.0x and the lowest at 5.0x, a spread of 16.0 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 8.0 of the 518.5 human-equivalent hours, or 2 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 157 minutes against 1,059 minutes of execution, a ratio of about 1 to 7. Supervisory leverage of 198.2x 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.