Skip to content

Instantly share code, notes, and snippets.

@brynary

brynary/fabro-effort.md

Last active March 20, 2026 18:26
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save brynary/7fc6c06fdf30eec47c481d3d6357befc to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save brynary/7fc6c06fdf30eec47c481d3d6357befc to your computer and use it in GitHub Desktop.
Download ZIP
Fabro Project Effort Analysis
Raw
fabro-effort.md

Fabro Project Effort Analysis

Timeline
Start date Feb 19, 2026
Latest commit Mar 20, 2026
Duration ~30 calendar days
Days with commits 30 / 30 (every single day)

Scale

Metric Value
Total commits 1,401
Rust crates 25
Current codebase ~188,000 lines
Rust ~155,000 lines
TypeScript ~21,000 lines
Total insertions 353,000
Total deletions 113,000
Net lines written ~240,000

Contributors

Author Commits Share
Bryan Helmkamp 1,357 97%
Fabro agent (automated) 909 fmt / verify / fixup / simplify
Bots & others 44 3%

AI Involvement

This is heavily AI-assisted development:

  • 92% of commits (1,292) carry a Co-Authored-By tag (Claude or similar).
  • 909 commits are fully automated Fabro agent runs (formatting, verification, simplification).
  • The remaining ~500 co-authored commits represent human-directed, AI-assisted work.

Work Pattern

Daily velocity

  • Average: 47 commits / day across 30 days
  • Peak day: 124 commits (Mar 10)

Weekly ramp

Week Commits
W08 (Feb 19-23) 39
W09 288
W10 480
W11 416
W12 (partial) 179

Hours (ET)

Block Commits
Morning (07-12) 424
Afternoon (12-18) 528
Evening (18-00) 314
Night (00-06) 136

Active every day of the week, including weekends (Mon: 266, Sat: 215, Sun: 251).

Effort Estimate

Human time

  • 30 consecutive days, working across all hours including nights and weekends.
  • Hourly distribution suggests ~12-16 hour days on heavy days.
  • Estimated 300-400 hours of human time (~10-13 hrs / day average).
  • Equivalent to 2-2.5 months of full-time work at 40 hrs / week.

What it would cost without AI

A project of this scope — 25 Rust crates, an API server, React frontend, CLI, multi-provider LLM client, sandbox abstraction, workflow engine, and integrations with GitHub, Slack, Linear, and MCP — built from scratch in 30 days is extraordinary.

Scenario Estimate
Solo senior engineer (no AI) 12-18 months
Small team (3-4 engineers, no AI) 4-6 months
Equivalent contractor cost (no AI) $400k-$750k at $200-250 / hr
Equivalent team cost (no AI) $300k-$500k (salaries + overhead, 4-6 months)

AI costs

  • ~1,300 AI-assisted commits generating and reviewing substantial code.
  • Estimated $2,000-$8,000 in API costs over 30 days, depending on context sizes and model tier.

Total actual cost

Component Low estimate High estimate
Human time (300-400 hrs @ $200-250 / hr) $60,000 $100,000
AI API costs $2,000 $8,000
Total $62,000 $108,000

Compared to the $400k-$750k it would cost without AI, the actual cost of $62k-$108k represents an 80-85% reduction — achieving the same output at roughly one-fifth to one-sixth the price.

Takeaway

This is a 5-10x productivity multiplier story. One developer, working intensely for 30 days with heavy AI assistance, produced what would traditionally require a small team working for several months. The 188,000-line codebase spanning Rust, TypeScript, and infrastructure is not trivial — it includes a workflow engine, LLM abstraction layer, multiple sandbox implementations, and a full API + web frontend.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment