Woman in a yellow coat talking on her phone while walking through a winter park

Voice AI

Voice AI vs. Chatbots: Why Talking Beats Typing for Reflection

Every journaling app now has a chatbot. Text conversations can help, but the research on voice, connection, and follow-through points somewhere else.

6 min readClaire Team

Every journaling app has a chatbot now. You type how you feel, it types something back, and the exchange becomes your entry. Claire has a text chat too, and it is useful. But the center of Claire is a phone call, and the reason is not nostalgia for the telephone. It is what the research says about how people open up, connect, and actually follow through.

Typing Is Composition. Talking Is Thinking.

A chatbot still presents you with a text box. It is a friendlier box, with a question above it, but you are still composing. You choose words, backspace, reconsider, and the editor in your head gets a vote on every sentence. Most people type around 40 words a minute and speak around 150, so a chat session also captures a fraction of what a conversation of the same length would.

Speech works differently. Bourdin and Fayol showed that writing taxes working memory more heavily than speaking, because you are managing spelling, grammar, and motor control alongside the thought itself. When you talk, that bandwidth goes to the thought. People routinely discover what they think by hearing themselves say it, which is the entire premise of think-aloud research and most of therapy. You do not need to know what you feel before you pick up the phone. You find out during the call.

Man walking a tree-lined path at sunset with a phone to his ear

What Voice Carries That Text Cannot

Amit Kumar and Nicholas Epley ran a series of experiments in which people reconnected with old friends by phone or by email. Beforehand, most expected the call to be more awkward and preferred to type. Afterward, the people who called reported feeling significantly more connected, and no more awkward, than the people who wrote. The mistake was systematic: people undervalued voice before using it.

In an earlier study with Juliana Schroeder and Michael Kardas, Epley found that hearing someone explain a view they disagreed with made listeners judge that person as more thoughtful and more human than reading the identical words. The voice, with its pauses, emphasis, and hesitation, revealed a mind that text concealed.

Those studies are about people talking to people, and we will not pretend an AI is your old friend. But the mechanism runs in both directions. When you speak, more of you shows up: the catch in your voice on a sentence you thought was neutral, the long pause before “it's fine.” A voice AI hears the “it's fine” and can ask what fine means today. A chatbot receives a tidy sentence you already edited.

You do not need to know what you feel before you pick up the phone. You find out during the call.

To Be Fair, Chatbots Help

In 2017 Kathleen Fitzpatrick and colleagues ran a randomized trial of Woebot, a text-based conversational agent delivering cognitive behavioral content, with 70 young adults over two weeks. The chatbot group showed a significant reduction in depression symptoms compared with a control group given an informational ebook. Text conversational agents can move the needle, and anyone who tells you otherwise is selling something.

So the question is not whether text works. It is what each medium is good at, and which one you will still be using in a month.

The Medium Decides Whether You Show Up

This is where the comparison stops being academic. A chatbot lives in an app, and an app waits. Baumel and colleagues measured real-world engagement across 93 mental health apps and found median 30-day retention of 3.3%. The apps were not bad. The design asked people to remember, open, and type, every day, with nothing initiating on their behalf.

A voice AI that calls you inverts that. The prompt is an incoming call; the first step is answering it; the journaling is the conversation that follows. You can do it while walking the dog or sitting in the car before work, which is when most people actually have five unclaimed minutes. Text requires a screen, two hands, and a place to sit. Voice requires none of those.

When Text Is the Better Choice

Sometimes you cannot talk. You are in an open office, or the house is finally quiet at 11 p.m., or the thing on your mind is something you are not ready to say out loud yet. Text is also better when you want to reread and revise, or to leave yourself a precise note rather than a reflection. That is why Claire keeps a text chat alongside the calls. It is there for the moments between conversations, not a replacement for them.

How Claire Uses Both

The daily call is the practice. Claire rings at the time you chose, walks you through your modules, asks follow-ups, and turns the conversation into a written entry with a transcript, a mood rating, and the audio so you can listen back. The chat is the supplement: a thought at lunch, a follow-up on something from the morning. One medium initiates the habit and carries the depth. The other catches what falls between.

Chatbots made reflection easier to start. Voice makes it easier to mean, and a call makes it easier to keep. If you have tried the text box and quietly stopped, the problem may not have been you. It may have been the box.

Ready to try voice journaling?

Start your free week with Claire and experience the difference of speaking your thoughts aloud.

Start Free Week

Continue Reading

Explore more insights about Voice AI and personal growth.

Back to All Articles