Interview preparation and write-ups with AI

Interviews are some of the most valuable content a membership can offer. A conversation with an experienced practitioner, a respected expert or a successful member brings in a voice that isn't yours and knowledge you don't have. But interviews take real work: researching the guest, planning questions, recording, transcribing and then turning a rambling hour into something readable.

AI can help at every stage except the conversation itself. It speeds up preparation, handles transcription and does much of the heavy lifting in the edit. What it can't do is listen, follow up on a surprising answer or make a guest feel comfortable. And because you're publishing someone else's words, the review step matters even more than usual.

Research the guest, then verify

Good questions come from knowing your guest. Eleanor runs an association for owners of historic homes and interviews a restoration specialist, architect or craftsperson each month. Before each interview, she gathers the guest's bio, a few articles they've written or talks they've given, and anything they've shared publicly about their work.

She then asks an assistant to help her see the shape of their expertise:

I'm interviewing [guest name and role] for my membership of [describe your members]. Here's what I know about them: [paste bio, article excerpts and notes]. Summarize their main areas of expertise and any views they hold that might be distinctive or debated. List topics they've already covered publicly, so I can avoid repeating them. Only use the material I've provided; don't add information about this person from anywhere else.

That last instruction matters. AI tools can invent plausible-sounding details about real people, including jobs they never held and books they never wrote. Research tools with citations, such as Perplexity, can help you find sources, but read the sources themselves before you rely on anything. Never put an unverified claim about a guest into your questions or your write-up.

Plan questions members care about

Once you understand the guest, use AI to widen your list of possible questions, then choose the best ones yourself.

My members are [describe them and what they struggle with]. Based on this guest's expertise, suggest 20 interview questions. Include: questions about specific practical decisions, questions about mistakes and what they learned, questions that invite a story, and a few that gently challenge a common view in the field. Make them open questions, not yes-or-no. For each, add one possible follow-up.

Eleanor picks eight to ten questions and adds two or three from member suggestions, which she collects in the community a week before. She puts the most important questions near the start, in case time runs out. The follow-ups are a safety net, not a script; the best moments usually come from following what the guest actually says.

Record with consent and transcribe

Tell your guest in advance how the interview will be used: recorded, transcribed with an AI tool, edited and published to members. Ask for their agreement in writing, and offer them the chance to review the write-up before it goes live. Rules about recording and publishing conversations vary, so check what applies to you with a qualified professional.

For transcription, tools such as Otter.ai and Descript, or the built-in transcription in many meeting tools, produce a workable first draft. Transcribing recordings with AI covers how to get cleaner results. Correct names, places and technical terms in the transcript before you do anything else, since those errors spread into every later step.

Edit the conversation into a readable piece

A raw transcript is hard to read. Decide on a format first: a Q&A that keeps the conversation's shape, or a feature article that weaves quotes into your own narrative. Q&A is faster and keeps the guest's voice; a feature takes longer but reads more smoothly.

Here's the transcript of my interview with [guest]: [paste transcript]. Edit it into a [Q&A or feature] of about [length] words for my members. Remove filler words, false starts and off-topic tangents. Tidy the guest's spoken language for readability, but don't change their meaning, add words they didn't say, or combine separate answers into one quote. Keep their distinctive phrases. Suggest subheadings. Mark any passage where you had to guess the meaning with [CHECK].

If you've already turned live sessions into articles, the process will feel familiar; turning live session recordings into articles covers similar editing choices.

Review with the guest before publishing

Read the edited piece against the recording, paying special attention to every direct quote. Tidying spoken language is fine; changing what someone meant is not. Resolve every [CHECK] marker by listening to the original.

Then send the draft to your guest for a review of their quotes and any facts about them. Give them a clear deadline and explain that you're checking accuracy, not inviting a rewrite. Most guests appreciate the chance, and it protects both of you from misquotes.

ChatGPT, Claude, Gemini and Microsoft Copilot can all handle the preparation and editing steps. Many owners try the editing prompt in two tools, since some stay closer to the speaker's actual words than others. For more on planning interviews as member content, see interviews as member content.

Your interview checklist

1. Gather your guest's public material and summarize their expertise, using only that material.
2. Generate a wide list of questions, then choose the best and add member suggestions.
3. Get the guest's written agreement to record, transcribe and publish.
4. Transcribe and correct names and terms first.
5. Edit into a Q&A or feature, keeping every quote faithful.
6. Check quotes against the recording and send the draft to the guest before publishing.