Skip to content

End-to-End Walkthrough

This guide walks through a complete example: taking a Q3 financial report and turning it into a short, narrated audio summary.

You have a 15-page Q3 Financial Report in PDF format. You want to generate a concise, narrated audio summary that stakeholders can listen to instead of reading the full report.


  1. Go to Projects and click New Project.
  2. Name it something descriptive, like Q3 Financial Report — Audio Summary.
  3. Click Create.

  1. Open the new project and go to the Files tab.
  2. Click Upload Files and select your PDF.
  3. The platform will automatically read and index the document in the background.
  4. Wait until the project status shows Ready before moving on.

  1. Open Studio with this project selected.
  2. Click New Audio.
  3. Set the options:
    • Format: Brief (single narrator, around 4 minutes)
    • Language: English
    • Length: Default
    • Steering Prompt: “Focus on revenue growth, EBITDA, and regional performance.”
  4. Click Generate Script.

The platform will search the document, identify the most relevant sections, and write a narration script. When it finishes, the script is ready to review.


Read through the generated script in the editor. Check for:

  • Accuracy — are the statistics and figures correct? Cross-reference with the Grounding Report if needed.
  • Flagged content — the Grounding Report will note any statements the AI couldn’t tie back to your document. Review these carefully.
  • Phrasing — adjust any wording that doesn’t match your preferred tone.

Make any edits directly in the script editor.


  1. Click Render Audio.
  2. Select a voice — for example, Zephyr works well for executive summaries.
  3. Leave the format as WAV for the highest quality output.
  4. Click Synthesise.

The platform will convert each line of the script into speech, combine the audio, and generate a cover image for the piece.


When the render job completes, your artifact is ready. From the artifact detail view you can download:

  • The audio file — your narrated overview.
  • The transcript — the plain-text version.
  • The grounding report — a record of every factual claim and its source in your document.
  • The cover image — an AI-generated thumbnail for the piece.

In a few minutes, without writing a single line of copy, you’ve produced a professional, grounded audio summary directly derived from your source document. Every claim in the audio can be traced back to a specific section of the original report.

Use this quick decision path when you are not sure where to start:

  • Need orientation or onboarding: begin with Getting Started, then return to this guide.
  • Need to complete one concrete task: use the step-by-step flow in this page first, then expand into linked feature pages.
  • Need governance or permissions context: cross-check relevant admin pages before changing team or policy settings.

If the page workflow does not match what you see in the product:

  • Confirm your role and project permissions. Some actions are hidden for non-admin or non-owner users.
  • Confirm you are in the correct project context before running generation or review actions.
  • Refresh your page after major workflow actions (approval, status updates, reruns) to ensure queue and status cards are current.

Why do I see fewer actions than another teammate?

Section titled “Why do I see fewer actions than another teammate?”

Role and project-level access controls can hide controls you do not have permission to use.

How do I know I completed the workflow correctly?

Section titled “How do I know I completed the workflow correctly?”

Use status indicators and queue transitions as checkpoints. A successful move to the next queue is the strongest signal.

When should I escalate instead of retrying?

Section titled “When should I escalate instead of retrying?”

Escalate when repeated retries fail with the same error or when policy restrictions block your intended action.