Skip to content

Interview Transcript to Feature Draft

Example prompt: "When I drop a transcript into our 'Interviews — Ready' Google Drive folder, read it and draft a 1,200-word feature in Google Docs. Open with a two-paragraph scene or hook drawn from the strongest on-the-record moment, follow with a nut graf that states what the piece is about and why now, then walk through the three or four main threads of the conversation with quoted passages handled as proper pull quotes. Keep the subject's voice intact and don't invent anything that isn't in the transcript. At the top of the doc, include a headline shortlist of five options, a 40-word standfirst, and a list of any factual claims that need checking. Save the draft to our 'Features — In Progress' folder and post a link to #features in Slack."

The Problem

Turning a long interview transcript into a first draft of a feature is one of the slower parts of long-form journalism. The raw material is already good, but the writer has to read the whole transcript, decide on an opening, find the three or four threads that structure the piece, pull the best quotes, and resist the temptation to paraphrase where a direct quote would be stronger. A competent first draft from a 90-minute interview typically takes a full writing day, which delays everything else on the week's list.

How GloriaMundo Solves It

We build a workflow triggered when a transcript lands in the 'Ready' Drive folder. An integration step reads the transcript. An LLM step identifies three or four main threads in the conversation and picks candidate opening moments grounded in the strongest on-the-record exchanges. A second LLM step drafts the feature to a set length, using direct quotes verbatim rather than paraphrase, and writes a nut graf, standfirst, and headline shortlist. A third LLM step lists any factual claims in the draft that need verification and flags any passages where the writer might want to check the original transcript. An integration step saves the draft to the 'Features — In Progress' Drive folder as a Google Doc. A final integration step posts a Slack message with the link. The Glass Box preview shows the draft in full before the doc is created, so the writer can rewind and regenerate with different opening choices if needed.

Example Workflow Steps

  1. Trigger (file event): New file appears in the 'Interviews — Ready' Google Drive folder.
  2. Step 1 (integration): Read the transcript contents.
  3. Step 2 (LLM): Identify the three or four main threads and shortlist candidate opening moments.
  4. Step 3 (LLM): Draft the feature to the target length, using direct quotes verbatim and honouring the interview's tone.
  5. Step 4 (LLM): Produce a five-headline shortlist, a 40-word standfirst, and a list of claims to fact-check.
  6. Step 5 (integration): Save the draft as a Google Doc in the 'Features — In Progress' folder.
  7. Step 6 (integration): Post a Slack message in #features with the document link and a short description.

Integrations Used

  • Google Drive — source 'Ready' folder and destination 'In Progress' folder
  • Google Docs — the draft output in a familiar editing surface
  • Slack — a single link to the draft, posted to the features channel

Who This Is For

Feature writers, magazine editors, and in-house long-form teams who work from recorded, transcribed interviews and want a well-structured first draft in front of them instead of a blank page.

Time & Cost Saved

A thoughtful first draft from a 90-minute interview typically takes a full working day for a writer with the material fresh in their head, and longer if the transcript has been sitting for a week. This workflow produces a structured first draft in a few minutes, leaving the writer to do the work only they can do: tighten the prose, cut where needed, and choose the headline. A realistic saving is four to five hours per feature, with the extra benefit that pull quotes are verbatim rather than remembered.