Skip to main content
AI MAY 20, 2026

Leverage Record: May 20, 2026

11 tasks. May 20, 2026 weighted to 54.5x leverage across 550.0 human-equivalent hours in 605 Claude-minutes. Supervisory leverage closed at 1269.2x.

11 tasks. May 20, 2026 weighted to 54.5x leverage across 550.0 human-equivalent hours in 605 Claude-minutes. Supervisory leverage closed at 1269.2x.

13.8 weeks of human-equivalent throughput in 10.1 hours of Claude wall-clock. The 202.1x ceiling came from a knowledge graph Phases 4-31 complete — REST route table + 20 Act-II inventions (heartbeat, lens, focus, predictor, capture, topography, resonance, prefetch, flame, oscilloscope,...; the 14.4x floor sat at Full an inference engine content audit + generate v2 lesson atoms for AWS Solutions Architect Pro (893/894 atoms, diagnosed and fixed max_tokens truncation bug).

About These Records
These time records capture personal project work done with Claude Code (Anthropic) only. They do not include work done with ChatGPT (OpenAI), Gemini (Google), Grok (xAI), or other models, all of which I use extensively. Client work is also excluded, despite being primarily Claude Code. The actual total AI-assisted output for any given day is substantially higher than what appears here.

Task Log

#TaskHuman Est.ClaudeSup.FactorSup. Factor
1a knowledge graph Phases 4-31 complete — REST route table + 20 Act-II inventions (heartbeat, lens, focus, predictor, capture, topography, resonance, prefetch, flame, oscilloscope, commitment, sentinel, topology, SLO tattoo, genome, spatial briefing, a CMS publisher, audience mirror, war room, ticker) + Act-II...320.0h95m1m202.1x19200.0x
2Fleet round 4: a newsletter platform multi-tenant newsletter ownership + migration, an accounting tool cash flow investing/financing wiring + invoice taxcode per-line calculation, a relationship CRM networkgraph_task end-to-end persistence + health endpoint fix, an analytics platform scheduled report dispat...42.0h75m2m33.6x1260.0x
3Fleet round 5: a knowledge base wiki-link resolution to real pages in same space, an infrastructure tool tag-keys-canonical custom in-process evaluator registry, a marketing platform lead scoring service writing to Contact.lead_score30.0h55m2m32.7x900.0x
4Fleet round 7: a CMS FR-013 content export ZIP endpoint, full doc-vs-implementation alignment audit across 21 backend tools (an observability platform/a defect tracker/an AI tool/an email platform/a portfolio browser/a gateway/a metrics tracker/a CMS/a calendar platform/a relationship CRM/a monitoring tool/a...20.0h40m2m30.0x600.0x
5Fleet round 3: an observability platform SLO budget action dispatch, a calendar platform in-process scheduler, a defect tracker activity WS broadcast, a marketing platform campaign step PATCH + EmailEditor save wiring, an audio tool @mention email dispatch40.0h80m2m30.0x1200.0x
6Fleet round 6: a task tracker FR-SHARE-020 notification service worker, an audio tool FR §3.12 built-in slash commands (/me /shrug /status /away /dnd /topic /archive /leave /remind)16.0h35m2m27.4x480.0x
7Fleet round 5c: an infrastructure tool governance.enforcementpolicies + .enforcementviolations ops + frontend rewire, an infrastructure tool expires-on-not-passed + expires-on-required-in-dev custom evaluators, an infrastructure tool IpSpacePage and AdvisorPage stale TODOs cleared18.0h40m2m27.0x540.0x
8Fleet feature implementation round 2: a marketing platform landingpage prompt builder, a calendar platform event attachments end-to-end, an observability platform alert.firing -> a notification service dispatch with publishafter_commit, a defect tracker @mention notifications28.0h65m2m25.9x840.0x
9Fleet round 5b: a CMS frontmatter TODO cleared, a marketing platform site list/settings frontend error display, an infrastructure tool StackDetailPage costs.by_stack wiring12.0h35m2m20.6x360.0x
10Phase 1 recommender starvation fix (lesson-first + goal-scoped saturation + weakgoalids surfacing) + SAP-C02 baseline scenario family (vacation, recert, convoy) + lessons-learned doc + validation sweep runner. 5 new commits (2 engine, 3 decoy). 9 new regression tests; full 5999-test suite green.18.0h60m4m18.0x270.0x
11Full an inference engine content audit + generate v2 lesson atoms for AWS Solutions Architect Pro (893/894 atoms, diagnosed and fixed max_tokens truncation bug)6.0h25m5m14.4x72.0x

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 day's leverage distribution matters more than the headline figure. The 202.1x ceiling came from a knowledge graph Phases 4-31 complete — REST route table + 20 Act-II inventions (heartbeat, lens, focus, predictor, capture, topography, resonance, prefetch, f...; the 14.4x floor was Full an inference engine content audit + generate v2 lesson atoms for AWS Solutions Architect Pro (893/894 atoms, diagnosed and fixed max_tokens truncation bug). Tasks at the top of the distribution share a shape: tightly-scoped specifications, clear success criteria, and minimal integration ambiguity. The AI doesn't need to discover anything new; it executes against an explicit target.

Tasks at the bottom run differently. They're either bounded by review-heavy work where every step gets verified, or they involve ambiguity that demands several rounds of trial and adjustment. The factor is real and informative, not a failure mode.

The supervisory leverage figure (1269.2x today) tracks something orthogonal to wall-clock leverage. It's the ratio of human-equivalent output to human prompt-writing time. It stays high even on lower-leverage days because supervisory minutes scale with task count, not with the human-hour estimate; a 20-minute task and a 4-hour task can both be specified in two minutes of human prompt-writing.

Across the 11 tasks, the day produced roughly 13.8 weeks of senior-engineer-equivalent throughput in 10.1 hours of model wall-clock. That ratio is the practical answer to the question of how much output a single operator can move per day when the model handles the execution and the operator handles the direction.