Skip to main content
AI10 July 2026

Coding and deployment

15 tasks. July 10, 2026 closed at 22.2x weighted leverage across 319.0 human-equivalent hours in 862 minutes of wall-clock time. Supervisory leverage came in at 236.3x.

That is 8.0 weeks of human-equivalent throughput in 14.4 hours. The ceiling was 54.0x; the floor was 8.6x. 15 of the 15 entries came from a single project.

Task Log

#TaskHuman Est.ClaudeSup.Factor
1AVIAN sim standardization: vendored amCharts5 (+Maps) offline, built the Sim SDK (shared shell/controls/theming + first-run coach-mark walkthrough + per-variable help + Guide modal), rebuilt Supply&Demand on…36.0h40m4m54.0x
2AVIAN simulations: ElevenLabs narration pipeline (baked 6 clips + rewired blood-flow to play them, no browser TTS) + built one reference sim per interaction type (guided_builder Circuit Lab, data_widget…40.0h45m3m53.3x
3Audit and review36.0h48m2m45.0x
4Documentation30.0h45m5m40.0x
5avian-[engine subsystem] remediation: verified stale-content failure is resolved + authored 5-phase remediation plan (all findings traced, Mermaid dep map, open decisions)5.0h10m2m30.0x
6ADR-0010 fast-boot Phase-0 foundation, 5 repos: 7-agent recon + read-only AWS verification + build-readiness report; engine (3 flags, WP0.2 boot instrumentation, WP2.3 readiness bifurcation, WP1.4a loader…80.0h170m4m28.2x
7AVIAN sims: built the macro AD–AS (Aggregate Demand & Supply) reference sim on the SDK+amCharts standard — AD/SRAS/LRAS curves, demand & supply shocks, output gaps, animated self-correction; registered +…5.0h13m1m23.1x
8AVIAN sim SDK: added animated graph transitions + draggable curves (SimChart.dragHandle) to both amCharts econ sims, a safe markdown renderer for tour/help/guide bubbles (fixed raw HTML), and removed the…4.0h12m2m20.0x
9AVIAN sims dev experience: container-side build-watch.sh + nginx sub_filter injecting livereload.js on every page — auto-reload on any file change (fixes Safari stale cache) + on-screen build-hash badge; no…2.0h10m2m12.0x
10G1 Finisher: SAA-C03 AWS content adjudication+staging (services/[engine subsystem]) -- 780-item confirm-judge, waive 989 false-positives, found regen_items capability gap + 3 platform bugs, staged test path16.0h88m15m10.9x
11G1 finisher: SAA-C03 independent adjudication (287/1276 confirmed, 80% FP discovery) + capability-gap diagnosis + 989 waivers16.0h88m15m10.9x
12Fleet campaign: deterministic sharding (CampaignConfig shards + [engine subsystem] wiring + plan generator rework, verified 4907 shards/1.14M items)9.0h50m6m10.8x
13[engine subsystem] self-recovery (lost-job auto-requeue + executor idle watchdog + infra liveness watchdog) plus supporting-services Postgres backup system + [engine subsystem]/[engine subsystem]-worker…28.0h165m12m10.2x
14ARMED Option-4 campaign ops-prep (4 items): hub_max_concurrency config wiring end-to-end, campaign heartbeat-TTL mislabel fix (async subprocess polling), fleet dispatch-plan generator + dispatcher (build…8.0h50m6m9.6x
15AVIAN sims fix: debugged (Playwright) + fixed the econ sims — HUD pointer-events swallowing chart clicks (drag blocked), amCharts snap vs a JS param-tween for animation, and mid-drag bullet recreation…4.0h28m2m8.6x

Aggregate Statistics

MetricValue
Total tasks15
Total human-equivalent hours319.0
Total Claude minutes862
Total supervisory minutes81
Total tokens10,795,000
Weighted average leverage factor22.2x
Weighted average supervisory leverage factor236.3x
Human-equivalent weeks8.0

Analysis

The highest factor of the day came in at 54.0x and the lowest at 8.6x, a spread of 6.3 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 36.0 of the 319.0 human-equivalent hours, or 11 percent of the day. No single task dominated the total, so the weighted average is representative.

Supervisory time was 81 minutes against 862 minutes of execution, a ratio of about 1 to 11. Supervisory leverage of 236.3x 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.