How GreenLight AI reads a story.
Four layers of analysis, distilled into one defensible verdict. Here's what happens between uploading a script and getting a score.
Genre & tone
The engine first places a story in context detecting genre and tonal signals from the text. A thriller and a romance are held to different standards, so this grounds everything that follows.
Commercial
Trained on 11,340+ real OTT titles, this layer forecasts predicted rating and the probability a project becomes a hit, reading production and market signals the way a seasoned buyer would.
Narrative
The heart of the engine. It scores character depth internal conflict, transformation, moral complexity, motivation and emotional arc — range, intensity, progression, and how memorably it resolves.
Audience
Finally it maps the story to audience clusters built from real engagement data, estimating who will actually show up and how strongly they'll connect.
One score. One clear call.
The four layers combine into a single GreenLight Score from 0–100 and one of three verdicts Invest, Pilot, or Pass — each with its reasoning attached. It's a data-driven indicator built to inform a decision, never to replace human judgment.
Why this matters
Most content loses money; a handful of hits carry the rest. If you can shift the odds even slightly backing more of the stories that land and fewer that don't the economics of an entire slate change. GreenLight AI exists to shift those odds, at machine scale, before the money is spent.
And because every script it reads and every minute audiences watch feeds back into the model, it gets sharper the longer BaazTube runs. The engine is the edge and the edge compounds.