Evidence first. Your rules second.
Is It Based? does not publish one universal rating. It builds a reusable evidence layer for each work, then applies the lens you choose. Two people can reach different fit results without changing the underlying facts.
We capture Grokipedia articles, critic reviews, blogs, and configured public discussions for movies and games. Wikipedia is not used. Each source is processed separately; raw snapshots stay private, while short verified excerpts link to the original source. Reddit, YouTube, Google reviews, and X comments are labeled as audience interpretations and cannot establish content findings on their own.
A pinned Lovable AI model reads one source document per call at temperature zero and returns schema-bound detections. Signals use four evidence states: detected, not detected, disputed, or insufficient. Prominence is low, medium, or high (1–3); explicit evidence of absence is 0. Silence means insufficient evidence. Stance is separately supportive, critical, mixed, or unclear and requires its own supporting quote. Every displayed quote must match the stored source text exactly or it is rejected.
The extractor records named people and organizations that the source explicitly ties to that title: actors, directors, writers, designers, developers, publishers, funders, and other verified roles. Professional history or attributable advocacy is allowed only when materially related to the work or a selected signal. We do not infer sensitive traits, collect unrelated private life, or assign a permanent ideology score to any person. An administrator checks each named claim and its citation before an analysis becomes public.
Create multiple private named lenses for movies, games, or both. Choose from 50 categorized criteria, then set prefer or avoid, importance, and an optional hard limit. You can evaluate presence, support, or criticism separately. A violated hard limit yields Avoid; an unknown hard limit withholds the verdict. Unknown preferences reduce evidence coverage. Switching lenses recalculates your fit immediately from the same evidence.
A source’s interpretation is not objective proof of a work’s intent. Conflicting sources remain visible, and stance may be unclear. No source can establish every nuanced claim. “Insufficient” is a first-class result, not proof that a signal is absent. This ten-title catalog is the first production release; sources, signals, and media types can expand after launch.