Joan Comadran

Dairector

AI-native film production platform, from narrative development through post-production, with a Claude-powered Script Doctor and dramatic-arc analysis.

Visit site →
Stack

Next.js 16

Year

2025

Role

Full-Stack AI Engineer

Field notes

I — III

Discovery

What we found before writing a line of code — three observations that defined what this project would, and wouldn’t, become.

01

What it is

Overview

Dairector is ADAUGE's film production management platform, built for collaborative visual storytelling, from narrative development through post-production.

It brings script, direction, shots, assets and review into a single act-based pipeline, with Claude woven in as a creative collaborator.

02

Why it matters

Challenge

AI-driven filmmaking is powerful but chaotic.

Narrative, direction, production and review live across scattered tools, chats and prompts, and holding a consistent creative vision across an entire project is hard when nothing is connected, and when a team of writers, directors, art leads and AI generators all touch the same story.

03

What we built

Solution

A single platform that structures the whole pipeline (tema → arco → guión → preproducción visual → producción → entregado) with role-based approval gates, versioned AI-generated assets, and timecoded screening reviews. Anthropic's Claude powers a Script Doctor that diagnoses the narrative and a Hype Graph that maps the dramatic temperature across the story's key moments.

Narrative & AI

A Script Doctor built on Claude

The narrative layer is a Tiptap rich-text editor with version history and approval gates, backed by an AI Script Doctor. It reads the script the way a story editor would, diagnosing the narrative and returning prioritized notes, and plots a Hype Graph, a dramatic-temperature arc across 20 key moments, so the team can see where the story sags before a single frame is generated.

Production

An act-based production pipeline

Production is modeled as act-based phases with shot tracking (lighting, camera and character metadata per shot), scene handoffs between team members, and multi-phase approval gates with role-based decision makers: guionista, dirección, director de arte, generador IA, project manager, ingeniero.

Assets & Review

Versioned assets, timecoded reviews

Every AI-generated asset (spaces, characters, first frames, video, audio, DaVinci timelines) is versioned. Screenings run with timecoded review comments organized by craft: etalonaje, guión, montaje, música, SFX, doblaje, so feedback lands exactly where it belongs on the timeline.