Wishdeal Factory mines fresh business ideas from twelve sources every day, scores them on eight axes, ships polished landing-page MVPs, generates founder personas and audio pitches, role-plays as skeptical buyers reading the homepages, and rewrites itself based on what works. Right now it has shipped 822 products, 10 at top-quality showcase tier. It runs around the clock with no human in the loop.
A bash loop on a single VPS wakes up every ten minutes, asks Claude to pick one valuable action from a 17-letter menu, and executes it. Each tick produces something concrete: a new MVP shell, a sub-page, a critique, a skeptic memo, an outreach draft, an audio pitch.
The studio learns from outcomes. Every six hours it audits its own catalog, identifies which categories produce showcase-tier products versus which ones consistently underperform, and writes that finding into the prompt of the next idea-mining run. Skeptic verdicts get clustered into recurring themes that feed back as Foreman demands across affected products. The result compounds.
The studio refuses to dilute. Every new build runs through a similarity guard that compares to the existing catalog. Same engine, same audience? Skipped as a duplicate. Same engine, different audience (e.g. a legal intake form vs. a construction intake form)? Built as a variant landing on the shared engine, with the relationship recorded.
Quality gets measured continuously. A 13-signal composite score (homepage depth, sub-pages, foreman reports, sales kit completeness, screenshots, skeptic memos, audio assets, founder persona, hero UI, outreach assets) sorts every product into Showcase, Solid, Rough, or Sketch tiers. The Director uses these tiers to decide where to invest next.
The 10 products at the top of the quality bar.
Every 822 products, sortable + filterable.
Faces and stories behind each product.
Real-named ICPs reading the homepages.
What the studio shipped today.
What the studio is bad at.
Hard-truth health metrics.
Live tick number + recent activity.