Provider routing workflow

Route each AI video shot to the provider that fits.

Video Loom turns model choice into a production plan. Start with the scene, decide which cloud or hosted provider fits the shot, keep prompts and references attached, then finish the cut on one timeline.

Open provider routing starter See how it works

Video Loom generation progress with provider-routed scene takes

Provider routing starter

Open the starter when a project needs different models for different jobs: performance shots, product proof, source-video restyles, transitions, social cutdowns, or abstract visual sections.

Map shot goals, aspect ratios, source media, and delivery formats before generation.
Decide where cloud video, hosted inference, image-to-video, and source-video routes fit.
Move selected takes into captions, audio sync, review, and export.

Model choice at the shot level

A performance closeup, an abstract lyric visual, a bridge transition, and a source-video restyle may each need a different model. Video Loom keeps that routing inside the project.

Per-scene provider and model overrides.
Fallback chains for configured adapters.
Provider status, usage estimates, and diagnostics in the workspace.

Continuity travels with the request

Provider calls can receive the scene prompt plus available references, source media, continuity cues, and previous-shot context.

Reference frames and continuity anchors.
Text, image, and source-video paths.
Take metadata for review and retry decisions.

Cloud and hosted providers in one project

Route across configured cloud video APIs and hosted inference for speed, capability coverage, cost-sensitive shots, and recoverable production workflows while preserving a single project and timeline.

Configured cloud and hosted provider adapters with per-scene routing.
Hosted model catalog plus fallback-aware generation.
Cloud persistence and recoverable jobs where supported.
Format-ready timeline exports from the same project.

Provider routing

Open the multi-provider workflow kit.

Start from a project brief, then turn it into scene prompts, provider routes, fallback notes, continuity references, and export-ready timeline work.

Open provider routing starter Read workflow questions

Questions

What teams ask before opening the app.

Does Video Loom require one AI video provider?

No. It is designed for configured provider routing, so teams can choose the best available model per scene.

Can I use multiple hosted providers in the same project?

Yes. Configured hosted providers can participate alongside cloud providers, with per-scene routing and fallback controls.

What happens if a provider fails mid-generation?

Video Loom supports fallback chains and recoverable jobs where the provider allows it, so a failed take can retry or route to another configured model without losing the storyboard.