21 tasks. April 15, 2026 closed at 37.1x weighted leverage across 783.5 human-equivalent hours in 1,266 minutes of wall-clock time. Supervisory leverage came in at 361.6x.
That is 19.6 weeks of human-equivalent throughput in 21.1 hours. The ceiling was 200.0x; the floor was 4.6x. 17 of the 21 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | avian-app-web: port 17 legacy pages (Lesson, Readiness, Trajectory, Curriculum, Scenario, Analytics, Governance, CrossDomain, MicroChallenge, Chat, LabConsole, Profile, ProviderMastery, 4 Competitive) +… | 100.0h | 30m | 2m | 200.0x |
| 2 | avian-app-web: full parity pass (engine REST client, 8 new pages including SessionPreview/Exam flow/InitializationFlow/StudentProfile, SiteKeyGate, VersionChecker, DashboardTour, Pyodide, competitive… | 200.0h | 70m | 3m | 171.4x |
| 3 | purchase-service Stripe rebuild: canonical 6-plan catalog, beta coupons (Certs [cost]/mo + [cost]/yr, duration forever), comp entitlements (friends/family lifetime + journalist 30-day), tokenized… | 90.0h | 60m | 12m | 90.0x |
| 4 | Documentation | 40.0h | 30m | 5m | 80.0x |
| 5 | Documentation | 80.0h | 80m | 5m | 60.0x |
| 6 | Debugging | 56.0h | 80m | 4m | 42.0x |
| 7 | Design and frontend | 24.0h | 35m | 3m | 41.1x |
| 8 | Add comp entitlements admin UI to avian-admin: API client, Comps page with filters/table/pagination, AddCompModal with presets, Revoke/Extend actions, real BillingTab with subscription+comps+payments,… | 24.0h | 35m | 8m | 41.1x |
| 9 | avian-app-web: wire real @avian/console-sim via iframe multi-page entry + drive Dashboard from real enrollment/autopilot stores (remove hardcoded demo data) | 6.0h | 10m | 1m | 36.0x |
| 10 | Coding | 60.0h | 120m | 8m | 30.0x |
| 11 | Implement tokenized accept-invite flow in auth-service: invitations table, Alembic migration, InvitationService, updated admin invite endpoint with comp_welcome path, GET/POST accept-invite API endpoints,… | 16.0h | 35m | 8m | 27.4x |
| 12 | avian-app-web: fix blank-page SPA nav (RouteFade mode=wait + Outlet trap); real lab manifest (2048 labs) in Labs page + correct labId routing to console-sim | 5.0h | 12m | 1m | 25.0x |
| 13 | avian-app-web: full module + real lab list visible to unenrolled on CourseDetail; Autopilot hero dark-mode contrast fix | 3.0h | 8m | 1m | 22.5x |
| 14 | Testing | 8.0h | 22m | 4m | 21.8x |
| 15 | Deployment | 20.0h | 90m | 10m | 13.3x |
| 16 | Wire entitlement-based enrollment gating into auth-service: purchase_client.py, domain_catalog.py, enrollment_service.py gating, API 403 NEEDS_UPGRADE response, 12 new tests | 6.0h | 28m | 10m | 12.9x |
| 17 | Infrastructure | 4.0h | 22m | 3m | 10.9x |
| 18 | Infrastructure | 3.0h | 22m | 4m | 8.2x |
| 19 | FR-083 auto-enroll on browse for comp users in auth-service enrollment_service | 3.0h | 22m | 5m | 8.2x |
| 20 | Deployment | 3.5h | 35m | 3m | 6.0x |
| 21 | Autopilot Phases 2-7 + P_pass fixes + entity persistence + 12 readiness-tracking iterations; validated monotonic readiness climb 0.376→0.559 (13 sim days, no regression) before UI-rebuild tabling | 32.0h | 420m | 30m | 4.6x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 21 |
| Total human-equivalent hours | 783.5 |
| Total Claude minutes | 1,266 |
| Total supervisory minutes | 130 |
| Total tokens | 7,615,000 |
| Weighted average leverage factor | 37.1x |
| Weighted average supervisory leverage factor | 361.6x |
| Human-equivalent weeks | 19.6 |
Analysis
The highest factor of the day came in at 200.0x and the lowest at 4.6x, a spread of 43.8 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 100.0 of the 783.5 human-equivalent hours, or 13 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 130 minutes against 1,266 minutes of execution, a ratio of about 1 to 10. Supervisory leverage of 361.6x 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.