KAHANI: An AI-Powered Cinematic Production Workspace and Pre-Production Platform

Abstract

Managing creative workflow asset continuity during early film pre-production presents significant difficulties for independent creators, student cohorts, and writers. Traditional general-purpose conversational Artificial Intelligence tools operate primarily as disconnected text chatbots that generate unstructured paragraphs without understanding cinematic grammar, structural screenplay logic, regional styling, visual lighting ratios, or production metadata. This paper detailed the architecture, development, and assessment of KAHANI (“Picture Abhi Baaki Hai Mere Dost”), an innovative, AI-powered cinematic pre-production workspace designed to transform abstract creative visions into structured, production-ready assets. KAHANI utilizes a client-server architecture integrating a component-driven React.js single-page application frontend with an event-driven, decoupled Python Flask micro-backend. System intelligence relies on structured prompt engineering layers and a localized Cinematic Language Selection engine that automatically shifts narrative pacing, emotional density, dialogue idioms, and frame metrics to emulate stylistic patterns such as Bollywood Mass, Malayalam Realism, Korean Thriller, and A24 Indie. Empirical evaluation validates that migrating operations to this centralized workspace reduces multi-department initial ideation loops from 4 to 5 hours to under 2 seconds (>99% acceleration), while eliminating data structural fragmentation, securing local client storage, and outputting production-ready documents via browser-native APIs.