Faster, Smarter, More Efficient: AI to Select the Best Film Scripts
In Russia, an AI-driven prescoring model is reshaping how film scripts are reviewed. The tool cuts evaluation time from 10 hours to just 10 minutes, enabling studios to process more submissions and make production decisions much faster.
Searching for the Best Stories
The team at KION, the online cinema under MTS Media, developed its own AI-based prescoring model for scripts. The solution was presented by Alexey Arefyev, Director of Product Portfolio Management at MTS Media, during the “New Season” festival for online streaming platforms.
“Each year we receive about 500 scripts, and previously each one took around 10 hours to review. In the end, only a handful made it to production—on average, 16 series and 5 feature films. Selection is a complex and time-consuming process, and it’s often hard to objectively evaluate a script,” Arefyev explained.
The team’s challenge was to reduce reading time, increase the volume of processed applications, set clear priorities, and accelerate production decisions. The logical step was to apply AI for automating routine but critical tasks.
The developers studied the stages of script evaluation and the underlying business processes, analyzed hundreds of scripts, broke them into parameters and components, and trained an AI algorithm to filter projects by specific criteria. Based on this, KION built its own model and launched an automated scoring system.
“Now it’s simple: we upload the script into a chatbot, the neural network runs it through the main criteria, and then generates a ready-made report for the editor—whether to ‘recommend’ or flag potential issues like a niche audience. Thanks to this scoring, producers can identify promising projects much faster,” Arefyev added.
How the Machine Works
The model analyzes a script’s logic—storyline consistency, absence of gaps, and proper sequencing of events. It also evaluates character depth, the complexity of their development arcs, and dramatic potential. The AI highlights opportunities for visual effects and measures technical difficulty, including camera angles, lighting, and special equipment needs.
Beyond technical analysis, the model includes the “KION DNA” metric, an internal tool that aligns projects with the platform’s values and strategic priorities. This ensures that the selected content matches KION’s overall positioning.
Instead of 10 hours, the system now needs only 10 minutes per script—delivering results comparable to those of a professional editor.
Global Interest
Netflix has long used AI tools for script analysis and popularity predictions. The Israeli startup Vault ML created a similar technology that forecasts box-office success based on structure, characters, and other elements.
The Russian solution is comparable in functionality but tailored specifically for selecting films for a single platform. It helps producers identify high-potential projects more efficiently, while also improving transparency by reducing subjective bias in reviews. This could attract more writers with innovative ideas.
Existing technologies reflect the growing demand for tools that accelerate data processing in creative industries. AI script scoring could become standard in Russia’s media sector within the next 2–3 years, and the model could be adapted for other platforms and evaluation frameworks.
Significantly, Russia has already developed a technology that may interest international streaming platforms and production companies, particularly in markets where AI-driven media tools are still underdeveloped.
