Portfolio Architecture

Four engines.
One studio.

Four production systems stand behind the ladder. They are not four products to sell — they are the supply chain for one. Each stands alone. Together they become the studio.

Explore the engines
Four glowing nodes connected by luminous data conduits in purple, blue, amber, and green

Shared disciplines

Every engine obeys four foundational rules.

I

Provenance as structure

Every artifact records where it came from, what produced it, and what decisions shaped it — from first generation through every revision.

II

Identity that survives regeneration

A line ID, a scene ID, a stem ID — each persists across rewrites, recompiles, and model upgrades. The production remembers what it is.

III

Review by exception

Surface only what needs human attention. Approved artifacts stay locked. Only flagged issues enter the review queue.

IV

Machine draft ≠ human ruling

The system proposes. The human approves, rejects, or overrides. That distinction is structural, never blurred.

The supply chain

Manuscript → Screenplay → Audio → Visuals → Release

Upstream

NarrativeStateEngine

Canon governance, world state, manuscript consistency

The franchise foundation. Maintains character registries, timeline continuity, world rules, and unresolved questions. Z3 formal verification. 10-stage pipeline, 120+ endpoints.

10-stage pipeline Z3 verified 120+ endpoints
Backbone

ScreenWright

Screenplay adaptation via Director/Writer/Critic over knowledge graph

Adapts prose into screenplay through ProtoScript IR, Fountain output, and a structured Ledger. 658+ tests, 78% coverage, 98% autonomous operation. The anchor economic metric: 45 of 80 human decisions auto-closed by ledger replay (~56% human labor reuse).

658+ tests 78% coverage 98% autonomy
Nearest deliverable

SoundWright

Narrative Audio Studio

Multi-voice production, table reads, stems, mixdown

Reuses reviewed speaker attribution from ScreenWright. Produces full-cast performances, stem-native sessions, auto-ducked mixes, and LUFS-compliant masters. ASVE six-entity binding times approved storyboards to audio. Forward door (and reverse door after launch). Provider-agnostic routing. 1.26M lines of code, 4,258 Python tests.

1.26M LOC 4,258 tests ASVE timing
Explore SoundWright →
Downstream

StoryBoard Engine

Directed storyboards, signed references, animatics

Turns a screenplay, Fountain or a ready Spine revision, into directed storyboards: the same cast and sets across every shot from references you sign, five looks, takes you compare and keep, and boards placed on the lines of the script. Follows a screenplay revision without starting over. Hands approved boards to SoundWright for timing. Runs as a self-hosted stack.

Signed references Five looks Revision-safe

Custody ladder

From hobbyist to studio platform.

Rung 1
Writers & Hobbyists — First Look
$15–$30/mo · Screenplay generation, optional audio preview
Cleared
Rung 2
Indie Authors — Locked Draft
$500–$2,500/title · Production-quality screenplay, audio optional
Cleared
Rung 3
Agents & Publishers — Batch Screening
Catalog evaluation, multi-title adaptation scoring
Gated
Rung 4
Producers — Production Draft
$10K–$50K/project · Full scene containment, revision tracking
Gated
Rung 5
Studios & Streamers — Full Platform
Cross-engine lineage, persistent production state, franchise tooling
Designed

The real moat

Amortized human judgment.

45 out of 80 human decisions on a ScreenWright script recompile were auto-closed by ledger replay — 56% human labor reuse. The moat is not gross margin per title. It is the compounding value of every human creative decision ever recorded in the system.

AI converters without structured governance suffer 339% extraction drift. CinemaWeave records, replays, and preserves every creative ruling.
Produce it with us