Patents
Patented AI technology
in the EU and United States
Original intellectual property developed at ScriptBook — covering automated AI methods for analysing, characterising, and generating scripted narratives. Invented and filed by Nadira Azermai.
Built from the ground up
ScriptBook's technology was not built on top of existing language models or borrowed AI infrastructure. It was developed from scratch — original research, original algorithms, original intellectual property.
The patents below cover the core methods that power ScriptBook's script analysis engine: how a screenplay is parsed, how its narrative factors are measured, how its potential success is predicted, and how new narratives can be generated from those same parameters.
Both patents were filed by Nadira Azermai as inventor, assigned to ScriptBook NV.
(EU · US)
Nadira Azermai
Patent № EP3340069A1 · Application EP16206566.8
Automated Characterization of Scripted Narratives
Abstract
What this patent covers
This patent covers the core technology behind ScriptBook's script analysis engine — the method by which an AI reads a screenplay and predicts its commercial and critical success, without any human intervention.
The innovation is a three-level approach: first, the screenplay is parsed into its structural elements (scenes, dialogues, actions, character names). From those elements, three types of narrative factors are extracted — topical factors (what the story is about), character factors (properties and traits of the characters), and novelty factors (how original the story is compared to others). Finally, a trained machine learning classifier combines all these factors to predict the story's likely success.
Key sub-methods covered include: scene-based topic modelling using Latent Dirichlet Allocation (LDA); similarity scoring between a screenplay and existing films based on audience viewing patterns; a "creativity factor" that objectively measures artistic originality from the script text alone; and character sentiment analysis that tracks emotional states throughout the narrative.
Key innovations
Patent № US20200334336 · Application US 16/389,198
Generation of Scripted Narratives
Abstract
What this patent covers
Where the European patent covers the analysis of scripted narratives, this US patent covers the generation of them. It is the intellectual foundation of DeepStory AI — the proprietary story generation platform N.D. Zerman built before the GPT era.
The patent covers a method by which the same narrative parameters used to analyse and evaluate screenplays (topical factors, character factors, novelty factors, audience factors) are inverted and used as inputs to a generative engine. Rather than extracting a script's fingerprint, the system uses a specified fingerprint to generate new narrative content that matches it.
This enables what ScriptBook called "story-awareness" in generative AI — the ability to produce screenplay content that is aware of genre, theme, character consistency, and narrative structure, rather than generating syntactically correct but narratively incoherent text.
Key innovations
Context
Why these patents matter
These patents represent original scientific work developed years before large language models became mainstream. ScriptBook's approach to script analysis was built on proprietary algorithms — not on top of existing AI infrastructure — giving it a fundamentally different basis to tools that simply applied GPT to screenplay-adjacent tasks.
The generation patent in particular anticipated the co-creation model that the AI industry would later converge on: not AI replacing writers, but AI as a story-aware collaborator that operates within parameters the human creator specifies. This was N.D. Zerman's position in 2019 — and remains the foundation of her thinking on mindful AI today.