Voice & Audio

Voice mode is a genuine two-way spoken conversation with an AI — you talk, it talks back, in real time, no typing involved. It's mobile-first because it shines exactly when your hands are busy: walking, driving, cooking, commuting.

The experience is different from typing in ways that matter. Speech is faster than typing for most people, and it's also less filtered — you tend to think out loud more freely when speaking than when carefully composing a message. That makes voice mode particularly good for brainstorming and processing, not just dictation.

A good way to start is to treat it like a conversation with a very patient colleague who has infinite time and no opinion about how long you ramble. The value comes from the back-and-forth, not from a single spoken question.

Dictation is speech-to-text: you talk, and your words appear as typed text, which you might then edit or send. Voice mode is a live conversation: the AI responds by speaking back, and the interaction continues.

Dictation wins when you know exactly what you want to say and just want to avoid typing it — a text message, a quick note, an email you've already composed in your head. Voice mode wins when you don't yet know what you want to say and need to think it through out loud with something responding to you.

Mixing these up leads to frustration: trying to dictate a complex plan in one breath is harder than talking it through conversationally, and using voice mode for a simple one-line note is slower than just dictating it.

One of the most underrated voice-mode uses is turning a walk into a structured thinking session. Instead of a vague meander through your thoughts, ask the AI to interview you — "ask me one question at a time about my priorities this week" — and let it hold the structure while you supply the content.

This works because most people think better when talking through a problem with someone (or something) asking follow-up questions, but rarely have a willing interviewer on demand. An AI in voice mode fills that role at 7am on a walk with nobody else awake.

The payoff comes at the end: ask for a written summary or plan based on everything you said. You arrive home with a structured document you built by talking, not typing.

Transcription takes any audio — a meeting, a lecture, a voice memo — and turns it into searchable, editable text. This alone is valuable: text can be skimmed, searched, and quoted in ways that audio can't.

The quality bar for everyday transcription is now high enough that most recordings, including ones with multiple speakers and some background noise, come out usable. This removes the old excuse for not recording things — you no longer need a stenographer or hours of manual typing to get a transcript.

The habit worth building is recording more often (with consent) precisely because transcription makes the recording actually useful afterward, rather than a file you'll never revisit.

A raw transcript is not the finish line — it's the raw material. The real value comes from asking for exactly what you need: a summary, a list of decisions, action items with owners, or specific quotes.

This is where most of the time savings live. A 45-minute meeting transcript might be 6,000 words; nobody wants to read that. But "summarize decisions, who volunteered for what, and open questions — 200 words" turns it into something you can read in 30 seconds and act on immediately.

It's worth asking for a couple of different cuts of the same transcript — a short summary for people who missed the meeting, and a detailed action-item list for people who were there and need to follow up.

The same transcription-then-summarize pattern applies to long-form audio you'd otherwise never get through — podcasts, recorded talks, audiobooks you're evaluating. You don't have to listen to two hours to know if something is worth your time.

A good approach is to ask for the key claims or takeaways with rough timestamps, then decide whether the full listen earns your commute or your evening. This turns "I'll listen to it eventually" (which usually means never) into an actual decision made in minutes.

It's not a replacement for listening when the content genuinely deserves your full attention — but for the pile of "might be useful" audio everyone accumulates, it's an efficient filter.

Voice and audio tools aren't just convenience features — for many people they're genuine accessibility tools. Hands-free dictation and voice mode help anyone who can't type easily, whether due to a temporary injury, a motor condition, or just being mid-task with both hands full.

On the other side, transcription helps people who have difficulty processing spoken audio in real time — turning a fast-talking meeting or a mumbled voicemail into text they can read at their own pace.

It's worth exploring these features even if you don't think of yourself as needing them; many people discover a workflow (like transcribing all voicemails automatically) that quietly removes a small daily friction they'd stopped noticing.