Ask your journal anything with Claude or ChatGPT

If you keep your journal in Monologue, you can connect it to Claude or ChatGPT and ask questions about things you've said: "what did I decide about the move?", "what have I been saying about work lately?" The assistant reads your transcripts, never your audio. It can't change or delete anything, and you can switch access off whenever you like, which deletes the copy kept for it.

This guide explains why you'd want that, exactly how it works, how to set it up, which questions get good answers, and what to check before you turn it on.

Why would you want to ask your journal questions?

Because most journals are write-only. You record, and record, and almost never look back. The entries pile up into something you'd value if you could find your way around it.

That's a real loss. In a set of four studies, people who documented ordinary moments in their lives underestimated how curious and interested they would later be to rediscover them (Zhang et al., 2014). The researchers linked this to overconfidence in our memory of everyday events. We think we'll remember. We don't.

Memory also bends. Once you know how something turned out, it tends to look more predictable than it did at the time, a well-documented effect called hindsight bias (Fischhoff, 1975). Your journal remembers what you actually thought before you knew. Being able to ask it is a way of checking yourself against your own record.

And some things are invisible one entry at a time. You might mention your manager in passing thirty times over three months without noticing the tone shifting. Search across everything you've said can surface that.

How does the connection work?

Here is exactly what happens, in plain terms.

• It's off by default. Nothing is shared with any assistant unless you turn it on yourself.
• Normally, your transcripts don't live on Monologue's servers. They live on your iPhone and in your own iCloud, end-to-end encrypted. While the connection is on, a copy of your transcripts is kept on Monologue's servers so your assistant can search them.
• It's transcripts, not audio. The assistant can search and read your transcripts along with their titles, dates and places. It never gets the recordings.
• It's read-only. The assistant can look, but it can't change or delete anything.
• You connect it yourself. You add Monologue to your own Claude or ChatGPT account and sign in with your Monologue account.
• Off whenever you like. Turn it off and connected assistants lose access right away, and the copy on Monologue's servers is deleted. You can also disconnect from the assistant's side.

One more thing worth saying clearly: what an assistant reads is then covered by that assistant's own privacy policy, not Monologue's. More on that below.

How do you set it up?

It takes a few minutes.

1. In Monologue, open Settings → Connect your AI and turn on Use with Claude or ChatGPT. You need to be signed in. A screen explains what the assistant can see and what stays on the server for it before anything is switched on.
2. Copy your connector link. Once it's on, the app shows a link and step-by-step instructions for Claude and for ChatGPT.
3. Add it in your assistant. Follow the steps the app shows for the assistant you use. Each one adds Monologue as a connector and asks you to sign in with your Monologue account.
4. Ask something. Start simple: "What have I been saying about work lately?"

Assistants change their menus and plan requirements over time, so treat the steps in the app as the current source of truth. Some features may need a paid plan on the assistant's side.

What should you ask your journal?

The best questions fall into a few families.

Decisions

• "What did I decide about the move, and what were my reasons at the time?"
• "When did I first mention wanting to leave my job?"
• "What did I expect to happen with the launch, before it happened?"

Patterns

• "What have I been saying about work lately?"
• "How has the way I talk about my sleep changed since June?"
• "On which days of the week do I talk about stress the most?"

People

• "What have I said about my sister this year?"
• "When did I last mention calling my grandmother?"

Ideas

• "List every idea I've had for the garden, with the dates."
• "What were the side-project ideas I got excited about and then dropped?"

Progress

• "What was I worried about in January, and did any of it happen?"
• "What have I said I'm grateful for most often?"

If you journal for a year, questions like these turn a review into a conversation. There's a fuller method in how to do a year-in-review with your journal, and a lighter one in the weekly review, spoken.

How do you get better answers?

A few habits make a big difference.

• Give it a time window. "In the last two months" or "since I started the new job" focuses the search.
• Name the topic in your own words. If you always call it "the house thing," say "the house thing."
• Ask for receipts. "Quote the relevant parts and give me the dates." Then you can check the source yourself.
• Ask for the other side. "What did I say that argues against this?" Assistants tend to agree with the framing you give them. Invite disagreement.
• Verify before you act. Assistants can misread, over-summarize or miss an entry. If an answer matters, open the entries it points to.
• Use your tags. If you tag your monologues, mention one: "only entries tagged Health." It can narrow things nicely.

What shouldn't AI do with your journal?

An assistant is good at finding and summarizing what you said. It is not a therapist, a doctor or an oracle. Don't ask it to diagnose you, and be wary of sweeping interpretations of your character. Treat it as a mirror held up to your own words, not as a verdict on them.

It also shouldn't replace the journaling. The value of talking things through comes mostly from doing it, not from analysis afterwards. We go further into this in AI journaling apps: what AI should and shouldn't do with your journal.

What should you check before turning it on?

Your journal is about as private as data gets. Before you connect anything, decide what you're comfortable with.

• The server copy. While the connection is on, a copy of your transcripts sits on Monologue's servers. If you only want to ask questions occasionally, you can turn it on, ask, and turn it off again, which deletes the copy.
• The assistant's data settings. Once an assistant reads your entries, its own policy applies. Both major assistants let you control whether your conversations can be used to improve their models: Anthropic describes its consumer data choices here, and OpenAI explains ChatGPT's data controls here. Check them before you ask about anything sensitive.
• Your chat history. Your questions and the assistant's answers stay in your chat history like any other conversation. Delete them there if you want them gone.

For a broader checklist on journal privacy, see how to choose a truly private journal app.

How this works in Monologue

Monologue is a voice journal for iPhone and Apple Watch. You talk, it transcribes, applies your own tags and gives each entry a cover photograph. By default your monologues live only on your iPhone and in your own iCloud, with transcripts, titles and places end-to-end encrypted, and Monologue's servers delete your audio once it's transcribed.

The Claude and ChatGPT connection is an optional extra on top of that: off until you turn it on in Settings → Connect your AI, read-only, transcripts not audio, and off again whenever you like. "How can I know what I think until I see what I say?" E.M. Forster asked. Now you can also ask what you said.

If you want a journal you can talk to, and then talk to about, get Monologue.