Batch Rendering and AI Agents

Cloud Rendering can start work from an editor, backend process, automation, or AI agent. Each request creates one tracked render job. This makes it possible to generate many videos from one template without using a browser for each output.

Common workflows

• Create regional or language variations of one campaign.
• Render personalised videos from approved customer data.
• Let an AI agent prepare a video plan and submit approved renders.
• Start renders from a scheduled workflow or content queue.
• Render from a dashboard without mounting the full editor.

Use one job for each output

Create a separate render request for each finished video. Store these values in your system:

• Your batch identifier.
• Your variation or campaign identifier.
• The RVE render identifier.
• The idempotency key.
• The current status.
• The output URL or error after completion.

Do not put personal data in the source name or external reference. Use your own internal identifiers.

Control concurrency

Do not submit an unlimited number of requests at once. Use a queue and a fixed worker count. Increase the worker count only after you test real videos and confirm the limit for your beta account.

async function runBatch<T>(
  items: T[],
  workerCount: number,
  renderOne: (item: T) => Promise<void>,
) {
  const queue = [...items];

  async function worker() {
    while (queue.length > 0) {
      const item = queue.shift();
      if (item) await renderOne(item);
    }
  }

  await Promise.all(
    Array.from({ length: workerCount }, () => worker()),
  );
}

Use stable idempotency keys

Give each logical output one stable idempotency key. If a network request fails before you receive a response, retry the same request with the same key. Use a new key only when you intend to create a new output.

AI-agent safety

An agent can prepare render variations, but your server must control submission.

• Validate the agent output against an allowlist or schema.
• Limit the number of renders in one action.
• Require approval for unusually large batches.
• Do not give the agent the Cloud Rendering API key.
• Record which workflow created each render.
• Keep customer data separate before you build the render payload.

Track progress

Save the render identifier returned by each request. Poll the progress endpoint until the state is done or error. Use the RVE Portal to review the complete batch, usage, output size, cost estimate, and failure details.