A Product Manager Workflow for User Interviews and Meeting Recordings

Product managers do not need another note-taking brand. They need yesterday's user interview and this morning's working session to become text they can quote, tag, and file before the next call. A transcript does that. A meeting bot, a research repository, and a slide summary are optional later steps.

Split the week into two recording types

• User interviews. One participant, a consent line, and a recording you will quote. The output is a cleaned transcript, not a cheerful summary that drops the hesitation.
• Internal working sessions. Decisions, owners, and open questions. A summary can be enough if someone who was in the room writes it. A transcript helps when you were facilitating and could not take notes.

Interviews that include unreleased product details or personal stories belong in the sensitive interview workflow. Do not drop those files into a shared meeting workspace by default.

Run one interview from recording to notes

1. Say what will happen to the recording before the first question. Include whether a bot will join. If the answer is no, record locally and skip the bot.
2. Save a file the participant could recognize: the meeting app's recording, or a phone memo if you met in person. The browser microphone inside Whisper Web records the laptop mic. It will not cleanly capture a remote participant unless their voice is playing into that mic.
3. Check the length. Under 20 minutes and 200 MB, transcribe on interview transcription in the browser. Over that, use Unlimited in the dashboard and treat the upload as a conscious choice.
4. Edit in passes, not all at once. First fix names and product terms. Then add speaker labels by hand, because the transcript will not do it. Then pull quotes into your research notes. Leave the rest in the file.
5. Export the version you will share. TXT or Word for the research doc. SRT or VTT only if a clip will be shown with captions. JSON if another tool on your side will parse timestamps. Generate caption files from subtitle generation when the clip is the deliverable.

What Whisper Web can and cannot do with the file

Use this only when the job is a recording you already have, or a microphone recording you make in the browser. Whisper Web is not a meeting bot, an enterprise meeting workspace, or a dictation layer that types into other apps.

• Free, in the browser: transcription stays on the device. Free files are limited to 200 MB and 20 minutes. A free dashboard batch accepts up to 5 files. Export the edited text as TXT, Word, SRT, VTT, or JSON.
• Unlimited, in the cloud: the published plan is US$20 per month, or US$120 per year. Files are uploaded for cloud transcription. The published limits are 10 hours or 5 GB per file, with batches of up to 50 files, saved history, and sync across devices. The pricing FAQ says those uploaded files are deleted after transcription. Cloud jobs need a network connection and start from the dashboard.
• Microphone, not system audio: the in-browser recorder asks for the microphone. It does not capture the other side of a Zoom, Meet, or Teams call. Record that audio in the meeting app or the operating system, then open the file.
• No automatic speaker names: the transcript is one continuous text. Add speaker names yourself while editing. The interview transcription page states this directly.
• First visit downloads a model: the free path needs a network connection the first time the speech model is fetched. The product FAQ says free transcription can continue offline after that download. This article does not add a new offline test, so try it on your own machine before you rely on a room with no network. Loading the website still needs a connection.

What to ignore in a "best notetaker for PMs" list

A tool that joins every call and writes action items can be right for a daily standup. It is a bad default for a discovery interview if the participant was not told, or if the note is a summary that cannot be traced to a sentence. Judge the tool by the artifact you can open next week:

• Can you get the raw words, not only the summary?
• Can you correct a name without losing the timestamp?
• Can you export a file your research doc can store, instead of a login to yet another archive?
• Does a 45-minute interview fit the free tier, or does it force an upload and a paid plan?

On Whisper Web, a 45-minute interview is over the free limit. Budget Unlimited, or transcribe a local excerpt first to see whether the draft is usable. If the interview is a formal record, read AI versus human transcription before you rely on the draft.

A light system for a team of one

Keep a folder per study. Put the recording, the exported transcript, and a one-page note in it. The note has the participant code, the date, five quotes with timestamps, and the decision the interview changed. That is enough to brief design without buying a research suite. Add a shared meeting product later if internal meetings, not interviews, are the bottleneck. The enterprise meeting-note guide is that later decision.

Frequently asked questions

Should every customer call have a bot?

No. Use a bot only when everyone on the call expects it and the team will actually read the shared note. For interviews, a file you control is easier to consent to and easier to delete.

Can I batch a week of interviews?

Yes, within the plan limits. Free dashboard batches stop at 5 files, and each file still has to fit 20 minutes and 200 MB. Unlimited batches accept up to 50 files, with the published per-file cap of 10 hours or 5 GB, and those files are uploaded.

Do I get speaker names automatically?

No. Add them while you edit. For a two-person interview that is a short extra pass. For a group workshop, budget more time or keep the transcript as an unlabeled record of what was said.