ovyn
← Case StudiesProduct XMethodology · Live
Methodology, made live
Type the intent. Watch the methodology.
A one-line ask goes into the engine. The model expands it into a structured PromptGraph — camera body, lens, film stock, lighting, palette, mood, forbidden visuals — before a single pixel is drawn. The image arrives afterward, anchored to those decisions.
Live · Product X brain · v18 voice rules · 18 forbidden phrases · 6 claims · 8 MLR prefs
The engine you’re about to demo runs against this brain. Generations go through both rails (model self-check + deterministic phrase scanner). Anything in the forbidden list below will not appear in any output — or the deterministic rail will flag it and the audit drawer will show you which rail caught it.
Quality-of-life claims attributed directly to Product XClaims that Product X cures or addresses the root cause of the indicationDisease-modifyingDisease modifyingsuperior to gene therapygene therapy will failtemporary solution until gene therapyStick used alone without Product X precedingnow approvednewly approved+8 more
Awaiting first intent
Type an intent above (or pick a preset) and hit Expand. The PromptGraph appears first — every camera, lens, and lighting decision visible. The Flux render comes second.