Skip to main content

Leverage Record: February 24, 2026

AI Time Record

About the author: I'm Charles Sieg, a cloud architect and platform engineer who builds apps, services, and infrastructure for Fortune 1000 clients through Vantalect. If your organization is rethinking its software strategy in the age of AI-assisted engineering, let's talk.

Daily accounting of what Claude Opus 4.6 built today. Twenty-eight tasks across seven projects. The day was dominated by standing up four product vertical websites with full AWS deployments, fixing diagram rendering issues across a large document set, and building out cloud infrastructure and backend services. This was the highest-volume day so far.

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.

The Numbers

# Task Human Est. Claude Leverage
1 Containerized deployment: Dockerfile, health endpoint, Terraform infrastructure, CI/CD pipeline 16 hours 8 min 120x
2 Four product vertical websites with AWS deployments, infrastructure subdomain support, and trademark compliance 80 hours 45 min 107x
3 Split monolith backend into API and Worker services 16 hours 12 min 80x
4 Product certification website (spec, components, 9 HTML pages) 16 hours 12 min 80x
5 Test preparation website (12 files, 5 exam subpages) 16 hours 12 min 80x
6 Enterprise product website (11 files, 5 industry detail pages) 24 hours 18 min 80x
7 Interactive adaptive learning frontend (16 files, React + Canvas) 16 hours 12 min 80x
8 Technology showcase pages (16 new, 13 modified across 5 sites) 16 hours 15 min 64x
9 Four domain specification documents with full validation (~306 lines each) 24 hours 25 min 58x
10 Cloud storage layer implementation (8 files, new subsystem, tests passing) 8 hours 10 min 48x
11 Remove proprietary disclosures from product website (15 deletions, 4 rewrites) 6 hours 8 min 45x
12 Fix figure rendering, margin overflow, and numeral collisions across 77 diagrams; regenerate 121 PDFs 24 hours 35 min 41x
13 Fix figure rendering and numeral collisions across 77 diagram files; regenerate 11 document packages 24 hours 35 min 41x
14 Dark mode styling, footer links, trademark symbols, and cross-site navigation across 5 websites 8 hours 12 min 40x
15 Documentation: OAuth setup guides, storage docs, API reference, local development 3 hours 5 min 36x
16 Static website infrastructure repository with full AWS deployment (S3, ACM, CloudFront, Route 53) 8 hours 15 min 32x
17 Exam coverage tracking: folder structure and READMEs for 110 certification exams across 6 providers 8 hours 15 min 32x
18 Conceptual article draft on engineering leverage measurement (~200 lines) 4 hours 8 min 30x
19 Fix four synthesis convergence bugs, build unified runner, resume two stalled processes to completion 16 hours 45 min 21x
20 Fix reference numeral collisions in two application diagrams 4 hours 12 min 20x
21 Legal pages (Terms of Service + Privacy Policy) for 5 websites 4 hours 12 min 20x
22 Draft page staging with visual DRAFT banners (5 files) 2 hours 6 min 20x
23 Live health status monitoring for dashboard with service documentation 3 hours 10 min 18x
24 End-to-end knowledge graph synthesis (375 nodes, 1,125 pairs, 93.6% validation accuracy) 8 hours 55 min 9x
25 Article updates: tracking prompt additions, chart styling, table font, draft badges 1.5 hours 12 min 8x
26 Template-based content partials; remove 30+ inline instances; fix breadcrumbs bug 3 hours 25 min 7x
27 Add randomized publish times to 30 articles and posts 0.5 hours 5 min 6x
28 Rename measurement terminology across all configuration, memory, and blog files 1 hour 10 min 6x

Aggregate Stats

Metric Value
Total tasks 28
Total human-equivalent hours 360
Total Claude minutes 494
Total tokens (approximate) 4,075,000
Weighted average leverage factor 43.7x

Analysis

Twenty-eight tasks is three times the volume of the previous day. The sheer breadth of work tells the story: four complete websites designed, built, and deployed; a backend split into microservices; Terraform infrastructure provisioned; 77 diagrams fixed twice across different document sets; domain specifications authored; and legal compliance pages written. No human team ships this in a day. Most of it would take a small team two to three weeks.

The product website cluster (tasks 2, 4, 5, 6) accounts for 136 human-equivalent hours. Each site involved a specification document, a shared component library, multiple HTML pages, and AWS deployment. The 80-107x leverage range on these tasks reflects work that is structurally repetitive but detail-intensive: each page needs correct content, navigation, styling, and deployment configuration.

The two diagram fix passes (tasks 12 and 13) total 48 human-equivalent hours. Fixing numeral collisions and margin overflow across 77 Mermaid diagrams, then regenerating output packages, is the kind of tedious cross-file work that exhausts human attention. An agent holds the full context and applies corrections mechanically.

The knowledge graph synthesis task (task 24) had the lowest leverage at 9x. Long-running synthesis processes with validation loops consume wall-clock time even for an AI agent. The 55 minutes of Claude time reflects iterative convergence rather than straightforward generation.

The weighted average of 43.7x means the day's 494 minutes of agent time replaced approximately nine months of focused solo engineering work. The total of 360 human-equivalent hours represents roughly nine 40-hour weeks.

Let's Build Something!

I help teams ship cloud infrastructure that actually works at scale. Whether you're modernizing a legacy platform, designing a multi-region architecture from scratch, or figuring out how AI fits into your engineering workflow, I've seen your problem before. Let me help.

Currently taking on select consulting engagements through Vantalect.