Tools, workflows, and economics: the agent stack (harnesses, skills, routines), the cost structure, and five worked case studies from idea to revenue.
Five builders, five receipts. Josh Pigford built NameSnag in about 12 hours and sold it for $15,000 eight days later (self-reported). Jack Friks runs Post Bridge to roughly $35K in monthly recurring revenue solo (self-reported) with AI agents doing his marketing. Nathan Wang runs 20–30+ agents 24/7 across trading, content, and operations. Ryan Carson ships about 10 pull requests a day solo. Robby Houston — non-technical — gave an agent a $100 budget and watched it build a paying community to $8.4K MRR in 13 days, with him as the "human in the loop" (all self-reported). This paper documents the agent stack they use (harnesses, skills, routines, loop frameworks), the actual cost structure (token economics down to cents per task), and what it costs to run one of these companies.
Summary drawn from the paper's executive summary (Human-N-Loop research series, October 2026 draft).
flagship credibility piece for AI Build Services ("we build these").
Sources named in the research: Pigford, Friks, Wang, Carson, Robby Houston, Sahil Lavingia, Alex Finn.