One self-taught operator, nine months, a B2B data pipeline built end to end — sourcing to self-monitoring — alongside AI agents that grew more capable over the same stretch. The same story, told four ways.
The human story. Why a man with no training beat the giants — David, the tweaker, and the dissolving of the last mile. (Illustrated.)
The narrative. The rot was in his own rules — and the fix, when it came, was to delete the machine that made it, not to perfect it.
The analysis. One self-taught operator builds a B2B data pipeline end to end — and what it shows about what one person can now do alone.
The case study. The decision, the Idea in Brief, the rebuilds, and discussion questions on human–AI co-evolution.
Every word and figure is drawn from 11,988 of the operator’s own messages across Cursor, Codex, Claude Code, and Cowork, Nov 2025–Jun 2026 — the same nine months in which the underlying AI models grew from single-turn assistants into fleets that run unattended. The man got better; so did the machines. They climbed together.