Skip to content

Cost analysis

One operator with AI tooling, against a modelled conventional team, over 117 days of one real project.

Read this first. The left column is measured: real sessions, real commits, real invoices. The right column is a model, priced at market salary plus a 1.4x loading for a team that was never hired. It is an estimate of an alternative, not a record of one.

The two columns

Scope: Campaign Brain Last 3 Months.

Measured2025-12-01 to 2026-03-26
Team size1
Commits27,939
Repositories134
Issues opened4,601
Issues closed3,096
Operator cost$60,000
AI cost$800
Total$60,800

The modelled team

10 people at $1,806,000 a year fully loaded. Salary figures are market rate; the 1.4x loading covers benefits, tax and overhead.

lower costhigher

Loaded cost by roleannual
Fully loaded annual cost
Data Engineer x2$378,000
Full-Stack Developer x2$336,000
ML/AI Engineer$217,000
Project Manager$203,000
DevOps Engineer$196,000
Biz/Workflow SME + UI Builder$182,000
Frontend Developer$161,000
QA Engineer$133,000

The gap

95.3% lower cost over the window, and 9 months to 1 daysin elapsed time. Both figures compare a measured column against a modelled one and inherit that model’s assumptions entirely.

The velocity multiplier the pipeline computes is 1,710x. It is not quoted as a headline here, because it divides by an active-day count of 1, and a ratio with a denominator that small is arithmetic rather than evidence.

What the work covered

Domain coverage0 to 100
Keyword coverage by domain
AI / ML79
Backend76
Security65
Systems59
Frontend50
IoT / Edge48
Data Engineering47
DevOps40

What the research says

Published findings on AI-assisted development, for calibration against the single project above.

SourceFinding
GitHub/Microsoft 202255% faster task completion
McKinsey 202320-45% productivity improvement
Google 202425%+ of new code AI-generated
BCG/Harvard 202340% higher quality output
Deloitte 202425-35% project cost savings

cost-model.json, generated Sep 6, 2026