I asked Claude Code to make me as much money as possible

An analytical walkthrough demonstrating four practical upgrades that turn Claude Code into a reliable business partner for product builders and agencies. The video builds a $9/month YouTube-to-LinkedIn-posts prototype while showing methods to force disagreement, verify outputs, manage conversation context, and run parallel workers. Each upgrade targets real failure modes—over-agreement, silent errors, context rot, and single-threaded bottlenecks—so outputs become higher quality, faster, and monetizable without rewriting the underlying model.

– Upgrade 1 — Roast (devil’s advocate): Spin up a council of personas (contrarian, expansionist, researcher, buyer, judge) to stress-test ideas, produce a clear verdict (greenlight/reshape/kill) and a cheap 48-hour test to validate market fit.

– Upgrade 2 — Verification loop: Force Claude to check its own work during build (Playwright screenshots, automated form submissions, edge-case tests) and iterate until artifacts are truly functional, not just superficially complete.

– Upgrade 3 — Context management: Prevent context rot by inspecting and compacting context, using session handoffs that summarize state and restore only essential information, and picking the right model for the task.

– Upgrade 4 — Parallelization & /goal: Remove the human bottleneck with sub-agents (each a clean Claude session) working in parallel and a /goal loop with a separate evaluator to run until objective completion, then verify outputs end-to-end.

Quotes:

Claude is tuned to make you feel productive. It is not tuned to make you money.

Researchers found AI models fail to push back about 88% of the time.

Stop being the bottleneck — use sub agents and /goal so stuff can run without you.