How to Build a Reported Feature Brief From Interview Notes With NotebookLM

The problem this solves and who it is for

This workflow is for editors, journalists, and feature writers working from messy reporting packets. You may have interview notes, transcript fragments, source documents, and a few links, but not yet a clean plan for the piece. The reporting exists. The story shape does not.

The goal is to create a feature brief that gives the writer or assigning editor a narrative spine, likely focus, missing reporting, and claim-check list before anyone starts drafting the full piece.

Prerequisites

• A Google account with access to NotebookLM
• Interview notes, transcript excerpts, or both
• Any supporting source documents such as reports, internal memos, or public pages
• A working story idea or assignment note
• Permission to use the material in your chosen tool

How to capture or gather the source material

1. Gather the current reporting packet in one folder.
2. If you have audio only, create a transcript first using your normal transcription workflow or device transcription feature.
3. Save transcript excerpts in a clean document if the raw transcript is long and messy.
4. Add one short assignment note that explains what the piece is supposed to answer.
5. Keep source names and dates in the documents so you do not lose attribution context.

Step-by-step workflow

1. Create one notebook per story. Do not mix different features in the same notebook.
2. Upload the reporting packet. Include the assignment note, transcript material, and any supporting documents.
3. Ask NotebookLM to map the factual terrain. First ask for major themes, strong story threads, and recurring points across the uploaded sources.
4. Ask for a narrative spine. Request two or three possible feature frames or structures and a note about which sources best support each one.
5. Ask for the reporting gaps. Have NotebookLM identify claims that need stronger support, missing context, missing voices, or chronology holes.
6. Ask for the final brief. Request a concise feature brief that includes story angle, narrative spine, must-verify claims, open questions, and editor notes.

Tool-specific instructions

Primary recommendation: NotebookLM

NotebookLM is the best fit because the work depends on keeping the brief anchored to your actual reporting. Google documents that NotebookLM centers on uploaded sources, source chat, and notes. That makes it well suited to feature planning from transcripts and supporting documents.

Practical setup:

• Upload the assignment note as a source, not just as a message.
• Keep transcript excerpts labeled by speaker where possible.
• Ask for missing reporting and weak support explicitly. Otherwise the tool may default to synthesis instead of scrutiny.

Alternative: ChatGPT Projects

ChatGPT Projects can handle this workflow if you upload the same documents. It is useful when you want file-backed conversation in a persistent workspace, but NotebookLM remains the cleaner source-first option for feature packets.

Alternative: Claude

Claude is a solid fallback for editorial memo work. If you use Claude, upload the documents together and ask for a strict brief format with reporting gaps called out separately.

Copy and paste prompt blocks tailored to the workflow

NotebookLM narrative mapping prompt

{
  "role": "assigning editor",
  "task": "analyze a reporting packet for feature development",
  "goal": "identify viable story frames grounded in the uploaded sources",
  "instructions": [
    "Use only the uploaded notebook sources.",
    "Identify the major themes, strongest scenes or examples, recurring tensions, and likely narrative threads.",
    "Suggest two or three possible feature frames.",
    "For each possible frame, note which uploaded sources best support it.",
    "Do not write the full article."
  ],
  "output_format": {
    "major_themes": [],
    "strong_examples_or_scenes": [],
    "recurring_tensions": [],
    "possible_feature_frames": []
  }
}

NotebookLM final feature brief prompt

{
  "role": "feature editor",
  "task": "write a reported feature brief from notebook sources",
  "goal": "produce a concise, source-grounded feature brief before drafting",
  "instructions": [
    "Use only the uploaded notebook sources and prior grounded analysis in this notebook.",
    "Write a brief that includes: likely story angle, narrative spine, strongest supporting material, claims that still need verification, open reporting holes, missing voices, chronology checks, and editor notes for the writer.",
    "Keep the result concise and useful."
  ],
  "output_format": {
    "likely_story_angle": "",
    "narrative_spine": [],
    "strongest_supporting_material": [],
    "claims_that_still_need_verification": [],
    "open_reporting_holes": [],
    "missing_voices": [],
    "chronology_checks": [],
    "editor_notes_for_writer": []
  }
}

Quality checks

• Confirm the brief is grounded in the uploaded reporting packet, not generalized feature-writing advice.
• Make sure missing reporting is clearly separated from well-supported material.
• Check that chronology questions are explicit when time order matters.
• Remove any claim that is not visibly supported by a source you uploaded.

Common failure modes and fixes

Failure mode: The brief sounds polished but vague.
Fix: Ask for source-backed examples and named reporting holes.

Failure mode: The tool smooths over contradictions.
Fix: Prompt specifically for conflicting claims or uncertain chronology.

Failure mode: The story frame is interesting but weakly supported.
Fix: Ask which frame has the strongest source support, not just the most dramatic arc.

Failure mode: The packet is too messy.
Fix: Trim the raw transcript and upload the most relevant excerpts first.

Sources Checked

• https://support.google.com/notebooklm/answer/16164461 (accessed 2026-03-25)
• https://support.google.com/notebooklm/answer/16215270 (accessed 2026-03-25)
• https://support.google.com/notebooklm/answer/16262519 (accessed 2026-03-25)
• https://help.openai.com/en/articles/10169521-projects-in-chatgpt (accessed 2026-03-25)
• https://support.anthropic.com/en/articles/8241126-uploading-files-to-claude (accessed 2026-03-25)

Quarterly Refresh Flag

Review by 2026-06-23 to confirm tool interfaces and supported file workflows still match the live products.