Claude video workflow

Build a Claude video workflow with review at every step.

Use Claude for a structured explainer script, ElevenLabs for narration and Remotion for editable scenes. Learn the workflow and choose a one-time Vedika kit.

  • Script → voice → scenes → render
  • Claude setup in all three kits
  • Provider access is separate
Explore one-time kits from ₹499

PRACTICAL WORKFLOW

What moves between the tools?

Use one project ID to keep the script, audio and rendered scenes together.

A concrete handoff: a weather app explainer

Ask for a scene where a weather app sends a city to a service and receives forecast data. Keep spoken words in narration, a diagram description in visualBrief, and short on-screen names in labels. Check the explanation before narration generation.

  • Claude → a structured plan in work/<project-id>/plan.json.
  • You → review the script and its factual references.
  • ElevenLabs → narration you listen to and approve.
  • Remotion → scenes timed to the actual audio.
Validate an existing plan before moving on

Replace my-explainer with the ID of a project you have already prepared. Validation checks the plan structure; it does not fact-check claims or call a script provider.

npm run kit -- validate my-explainer
  • If you change narration after approval, review the updated script again.
  • Keep the approved recording and the scene timings in agreement.
01

Give each tool a clear job

Claude helps prepare structured scripts and scene instructions. ElevenLabs supplies narration through your own configured account. Your Remotion project composes the visuals and produces the render. A script response is one stage of production, not a finished video. The kit provides project materials and instructions for connecting these stages.

02

Start with a brief the script can follow

Decide the audience, language, approximate duration and the one idea the viewer should understand. Supply reliable references for factual claims. Keep narration separate from visual directions so production instructions do not become spoken audio.

Topic: How an API request works
Audience: First-time learners
Language: Hindi / Hinglish
Target length: About 60 seconds
Scene: A phone sends a request to a service
Voiceover: The words the viewer should hear
Visual: A request travels from phone to server
03

Review the script before paying for narration

Read the complete script, check the source of each technical claim and remove visual directions from spoken lines. If you already have an approved script, retain it as your source of truth and request only the edits you need. Shorten or expand the wording after listening to an audition; a target duration alone does not fix speaking speed.

04

Build scenes around the approved voice

Choose and audition a suitable voice in your ElevenLabs account, then prepare the final narration. Use the actual audio length to set scene timing. Collect licensed assets or create diagrams, and inspect generated images for accuracy. Preview text, transitions and the full audio track before the final render. Creative and factual review stay with you.

05

Match the kit to your production process

Starter (₹499 once) gives you the Claude, ElevenLabs and local Remotion workflow. Pro (₹1,499 once) adds OpenAI as another script provider and Google AI image workflow materials. Max (₹3,499 once) adds advanced Claude production setup, the complete project blueprint, 1080p / 60 fps presets and AWS guidance. Model access depends on your provider account; no Claude subscription or API credits are bundled.

06

Keep configuration private and spending deliberate

Use the supplied .env.example as a placeholder template and keep real keys in your private local .env. Do not add that file to public source control or a shared Drive folder. Review your configured models and spending limits before paid generation. Customer-owned accounts mean provider charges and limits remain yours to manage.

Create with clarity

Understand the workflow before you choose.

Explore editable source, local setup, narration and the separate costs of API usage.