17 tasks. April 6, 2026 closed at 67.1x weighted leverage across 592.0 human-equivalent hours in 529 minutes of wall-clock time. Supervisory leverage came in at 455.4x.
That is 14.8 weeks of human-equivalent throughput in 8.8 hours. The ceiling was 240.0x; the floor was 15.0x. 8 of the 17 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | Documentation | 120.0h | 30m | 3m | 240.0x |
| 2 | Convert accelastudy.ai from static HTML to Vite React SPA with OIDC auth, domain catalog, Stripe pricing, enrollment flow, and full content port of 10+ pages | 120.0h | 45m | 5m | 160.0x |
| 3 | Design and frontend | 32.0h | 14m | 5m | 137.1x |
| 4 | Design and build [unreleased product] Certification system: 10 cert specs, JSON schema, DB models (2 tables), backend API (5 endpoints, service layer, Pydantic schemas), [unreleased product] frontend (catalog… | 80.0h | 45m | 10m | 106.7x |
| 5 | avian-admin observability backend: FastAPI service with health poller, incident detector, snapshot collector, S3 archival, 11 REST endpoints, 5 DB models, Alembic migration, and 6-tool MCP server | 40.0h | 25m | 5m | 96.0x |
| 6 | Design and frontend | 16.0h | 15m | 5m | 64.0x |
| 7 | Cloud CMS Phases 6-7: full frontend rebuild (Vite, auth, WebSocket, content editor), 45 exhaustive tests, mypy strict, Phase 8 deployment plan | 40.0h | 55m | 3m | 43.6x |
| 8 | avian-app-web refactor phases 1-4: remove pre-auth phases, require auth, repurpose Library as enrollment picker, URL param routing, env var feature flag | 12.0h | 18m | 8m | 40.0x |
| 9 | Refactor avian-app-web to post-auth dashboard app — remove pre-auth phases, add enrollment picker, URL param routing | 16.0h | 25m | 5m | 38.4x |
| 10 | Cloud CMS Phase 1-2: database models, schemas, Alembic, API routes, services, event bus, 95 tests | 24.0h | 45m | 5m | 32.0x |
| 11 | Port how-it-works.html and technology.html from static HTML to React TSX components with CSS | 3.0h | 6m | 3m | 30.0x |
| 12 | Cloud CMS Phases 3-5: build pipeline with materialization, content import tool, MCP server with 19 tools, 59 new tests | 32.0h | 65m | 3m | 29.5x |
| 13 | Testing | 16.0h | 35m | 5m | 27.4x |
| 14 | Testing | 20.0h | 45m | 5m | 26.7x |
| 15 | Coding | 3.0h | 8m | 3m | 22.5x |
| 16 | Deployment | 16.0h | 45m | 2m | 21.3x |
| 17 | Coding | 2.0h | 8m | 3m | 15.0x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 17 |
| Total human-equivalent hours | 592.0 |
| Total Claude minutes | 529 |
| Total supervisory minutes | 78 |
| Total tokens | 3,456,000 |
| Weighted average leverage factor | 67.1x |
| Weighted average supervisory leverage factor | 455.4x |
| Human-equivalent weeks | 14.8 |
Analysis
The highest factor of the day came in at 240.0x and the lowest at 15.0x, 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 120.0 of the 592.0 human-equivalent hours, or 20 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 78 minutes against 529 minutes of execution, a ratio of about 1 to 7. Supervisory leverage of 455.4x 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.