26 tasks. May 1, 2026 closed at 26.8x weighted leverage across 521.5 human-equivalent hours in 1,167 minutes of wall-clock time. Supervisory leverage came in at 284.5x.
That is 13.0 weeks of human-equivalent throughput in 19.4 hours. The ceiling was 69.5x; the floor was 5.1x. 22 of the 26 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | Testing | 44.0h | 38m | 4m | 69.5x |
| 2 | avian-automations: scaffold 2 containerized Lambda repos (data-export + account-deletion) + new terraform stack (ECR + Lambda + EventBridge bus + rules + S3 + IAM + CodePipeline x2) | 24.0h | 22m | 3m | 65.5x |
| 3 | Testing | 80.0h | 75m | 3m | 64.0x |
| 4 | AccelaStudy.ai site-wide updates: hide social proof + [unreleased product], merge comparison table from /certs/ to home, rewrite pricing copy, link accessibility bubbles, restructure courses page, filter labs… | 18.0h | 22m | 8m | 49.1x |
| 5 | Build avian-api Phase 0 + Phase 1 + first proxy slice + frontend EventBus + purchase-service event emission. New core/avian-api repo with FastAPI app, JWT/JWKS, Valkey pub/sub, SSE endpoint with Last-Event-ID… | 120.0h | 165m | 15m | 43.6x |
| 6 | Outstanding cleanup post-cutover: trim api. from engine listener rule (engine. only), backport ECS update IAM policy from CLI into avian-terraform with terraform import, move subscription_groups.json data… | 12.0h | 22m | 1m | 32.7x |
| 7 | avian-automations end-to-end deploy: 2 GitHub repos created and pushed, terraform stack applied (ECR + Lambda x2 + EventBridge bus/rules + S3 + IAM + CodePipeline x2), warm-invoke asyncpg pool bug found and… | 16.0h | 30m | 3m | 32.0x |
| 8 | Built competitive multiplayer feature on engine: full state machine (lobby/countdown/question/reveal/complete), in-memory CompetitiveStore, FFA + teams modes, scoring with point decay, REST endpoints… | 24.0h | 50m | 1m | 28.8x |
| 9 | auth-service emails: welcome on first verify + welcome on first social login + account-closed at close + new account-deleted template + APP_URL setting + 4 tests | 4.0h | 9m | 1m | 26.7x |
| 10 | AccelaStudy.ai polish round 2: courses page 3-wide with bulleted clean_titles for live+coming, drop Test Prep group, hide pricing testimonials, rename Other Products to AccelaStudy AI - X (Certs first),… | 6.0h | 14m | 4m | 25.7x |
| 11 | AVIAN Code Sandbox Phases 1-2-3: VirtualFileSystem, IDB cache, Zustand store, EditorHost (Monaco multi-tab), TerminalHost (xterm.js + 5 command resolvers: shell/git/npm/pip/validators-stub), ActivityBar,… | 35.0h | 90m | 15m | 23.3x |
| 12 | Phase 4 AVIAN Code Sandbox: S3 persistence — Sign-URL Lambda (JWT auth, quota, ETag conflict), expiry-warning Lambda, S3 bucket + Terraform, S3SyncEngine (debounce, hydrate, retry, conflict), IDB schema v2… | 24.0h | 62m | 7m | 23.2x |
| 13 | avian-api LIVE CUTOVER: terraform apply (25 AWS resources, ECS Fargate cluster + task + service + ALB tg + Route 53 + SSM + CodePipeline + IAM); 3 pipeline iterations to debug ruff config + coverage gate +… | 24.0h | 65m | 8m | 22.2x |
| 14 | avian-api side-effects dispatcher: bounded retry with exponential backoff + Redis dead-letter queue + ops inspection route + 6 unit tests; unblocks Item 4 Phase B | 7.0h | 25m | 1m | 16.8x |
| 15 | Batch 2 lab strict-pass: closed out 76-lab Top-100 push (resume after crash). 62 free wins from prior dashboard fixes; 14 hand-tuned (status case drift, prop-key filter mismatches, missing select testIds,… | 28.0h | 110m | 4m | 15.3x |
| 16 | Strategy phase 7 prod deploy: monitor CodePipeline build to completion, smoke-test admin endpoints over public engine domain, SSM port-forward to RDS, verify alembic head=005 includes… | 3.0h | 13m | 1m | 13.8x |
| 17 | Item 4 Phase B (retire engine direct PATCH; gateway delta=0->1; delete dead auth_client module + tests; trim auth-service _ALLOWED_CALLERS) + competitive multiplayer wired through gateway (REST proxy via… | 12.0h | 60m | 2m | 12.0x |
| 18 | Fix Anthropic API temperature-deprecation 400 in Origin LLM client (catch + drop param + latch for run); normalize 30 ISACA spec schemas (cost_usd dict→float, passing_score string→int with type,… | 8.0h | 40m | 5m | 12.0x |
| 19 | Infrastructure | 1.5h | 8m | 1m | 11.2x |
| 20 | wire CohortAnalytics By Plan + By Activity Level via cross-service in-process aggregation in admin-service (joins accelastudy-auth users + purchase-service subs); shared retention math | 4.0h | 25m | 2m | 9.6x |
| 21 | Three-repo cleanup sweep: avian-engine (delete tts.py + 5 test classes + legacy evidence route, tighten engine_context exception logging, README revert + count refresh, root-files allowlist), avian-app-web… | 10.0h | 70m | 3m | 8.6x |
| 22 | Fix avian-api gateway dropping user JWT on auth-service proxy calls (forward_auth plumbing through UpstreamClient + 19 client methods + 2 route files + regression test) | 3.5h | 25m | 6m | 8.4x |
| 23 | Wire prod REDIS_URL across admin/auth/purchase/notification services: 4 SSM params + 4 buildspec patches + 4 pipeline redeploys; root-cause for silently-dead events.* fan-out (publishers + admin-service… | 4.0h | 30m | 1m | 8.0x |
| 24 | Rewrite avian-app-web CLAUDE.md and README.md to reflect current React Router 7 + gateway architecture | 1.5h | 12m | 5m | 7.5x |
| 25 | fix Users tab (renkara-auth alembic up to head), Portfolio tab (admin-service ENGINE_ADMIN_KEY), Logins-by-Method dedupe; wire CohortAnalytics signup-month with new endpoint; revealed engine routes to… | 5.0h | 50m | 4m | 6.0x |
| 26 | Deployment | 3.0h | 35m | 2m | 5.1x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 26 |
| Total human-equivalent hours | 521.5 |
| Total Claude minutes | 1,167 |
| Total supervisory minutes | 110 |
| Total tokens | 5,978,000 |
| Weighted average leverage factor | 26.8x |
| Weighted average supervisory leverage factor | 284.5x |
| Human-equivalent weeks | 13.0 |
Analysis
The highest factor of the day came in at 69.5x and the lowest at 5.1x, a spread of 13.5 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 44.0 of the 521.5 human-equivalent hours, or 8 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 110 minutes against 1,167 minutes of execution, a ratio of about 1 to 11. Supervisory leverage of 284.5x 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.