Intelligence

Ra reads your work and reads what works. Every asset gets structured into memory. Every video feeds back: what converted, what held, what drove clicks. Ra knows what performs for your audience.

Intelligence is not a feature. It is a decision-making layer that applies judgment across many situations. “Add subtitles” is a feature. “Understand the emotional pacing of this scene and style the subtitles to match the tone while maintaining readability on mobile” is an intelligence. CS ships five of them, each a fat skill sitting under Ra’s thin harness.

The slogan

The Graph is the flywheel; Ra is the turbine.

The five Intelligences

Each Intelligence is a specialized layer that makes creative decisions in one domain. Ra routes the brief to the right Intelligence, the Intelligence applies judgment, deterministic tools execute. Fat skills, thin harness.

Intelligence	What it does
Media Intelligence	Understands media formats, metadata, optimal dimensions, compression, and cross-platform requirements. Powers Asset Manager’s smart organization and platform-native preview generation. What’s in the library, what belongs to whom, where things live.
Cinematic Intelligence	Understands film grammar: shot composition, scene transitions, narrative arc, character consistency, audio-visual sync. Powers AI Movie Maker’s script-to-video pipeline. Shots, beats, pacing, the film-grammar decisions a director makes.
Viral Intelligence	Understands what performs on each platform. A/B testing, hook optimization, posting time optimization, audience behavior patterns. Automated content optimization, tuned to what converts for your audience, not the median.
Broadcast Intelligence	Manages 24/7 streaming channels. Automated scheduling, live content rotation, audience engagement monitoring. Long-tail production at stream pacing.
AdX Intelligence	Brand deal matching. Understands creator profile, audience demographics, brand requirements. Marketplace operations and ad-unit optimization, campaign-specific craft.

Media and Cinematic Intelligence ship with V1.0. Viral, Broadcast, and AdX ship in sequence as the Graph matures and the signal density justifies specialization.

The conversion signal flywheel

Model providers (Veo, Kling, Runway, Pika, and others) operate at the generation layer. They see a prompt and return pixels. They never see what happens next:

• Which hooks held past the 3-second mark
• Which CTAs converted
• Which thumbnails earned impressions
• Which stories drove subscriber growth
• Which retakes the creator requested, an implicit quality signal
• Which final cuts shipped and which were scrapped

CS sees all of it. The loop:

1. Creator renders a story through Ra.
2. Story publishes to YouTube, TikTok, Instagram.
3. Performance data flows back: retention curves, click-through, conversions.
4. Data enriches the creator’s Director Memory Graph.
5. Ra’s next routing decisions optimize for what actually works for that creator’s audience, not just which model produces the prettiest frame.
6. Next story performs better, more data, tighter optimization.

Over time, Ra does not just route for quality. Ra routes for outcomes. “Use Seedance for this ambient shot because Seedance grain performs 12% better on retention for your audience’s demographic.” No model provider can make that call.

Ra is the turbine, the Graph is the flywheel

Ra’s routing logic is a mechanism, it makes the product work. The flywheel (Graph plus conversion signals) is the defensibility. Ra is explicitly positioned as the turbine, not the flywheel.

Why this is a moat

The downstream performance data no model provider ever sees is exclusive, and it compounds Ra’s routing decisions toward outcomes, not just visual quality. Over scale, the signal sharpens:

Scale	State
Pre-seed	Weak, not enough data to differentiate routing.
1K creators	Signal emerging, aggregate patterns visible.
10K creators	Defensible, outcome-optimized routing meaningfully outperforms quality-only routing.
100K creators	Unkillable, the dataset is a proprietary advantage no competitor can bootstrap.

Every story shipped through Ra enriches the Graph. Every enrichment tightens the next render. The moat is not the models, models commoditize. The moat is the closed loop between render and outcome that only a studio running end-to-end can see.

Works with

Memory

The Director Memory Graph is where Intelligence writes what it learns. Voice, cast, brand, outcome signals, all stored per-storyteller. Explore Memory →

Orchestration

Intelligence decides what should happen. Orchestration routes the fifteen-model stack to make it happen. Explore Orchestration →

Pillars overview

How the three pillars loop together on every brief. Pillars overview →

Meet Ra

The director with taste sitting above all five Intelligences. Meet Ra →