Irdora user guide

Every step from the first recording to the exported file, in the order you meet them. Screens and buttons are named as they appear in the app.

Last updated: 11 October 2026

Before you start

Summaries, action items, mind maps and answers need Apple Intelligence turned on, which runs on an iPhone 15 Pro or newer, an iPad with an M1 chip or later, or an iPad mini with A17 Pro, in a supported region and language. Recording, transcripts and speakers work without it, and the purchase screen says so before you buy. If Apple Intelligence is off or still getting ready, the app says that, and a recording made meanwhile is summarized once it is on, if Apple Intelligence supports its language.

The first time you use Irdora it prepares its speech model, which it downloads from Apple. It is a large download, so use Wi-Fi if you can. After that, Irdora connects only to the App Store, for purchases, and to iCloud, if you turn on sync.

Irdora Pro is required to use the app. It is a lifetime one-time purchase, a yearly subscription with a 3-day free trial, or a weekly subscription, and the App Store shows each price before you pay. When you first set up Irdora it also asks whether to keep your recordings on this device or to turn on iCloud sync. If a plan ends, your recordings stay on your device: you can still read, read aloud, export, share and delete them, and play them on the device that recorded them. Recording and new summaries need Irdora Pro.

Recording

Tap Record and have the conversation. Before your first recording Irdora shows a note, and Start recording on it carries on: recording other people may need their permission where you live, so let them know before you start. iOS asks to use the microphone the first time.

Irdora writes the audio to storage as you speak, and recording carries on when you lock the iPhone or switch to another app. A Live Activity on the Lock Screen shows the time and has a Stop button. You can also start a recording from the Action Button, a widget or Siri. iOS lets an app start recording only while it is in front, so Irdora opens first and then records.

When you stop, Irdora saves the recording and starts work on it. You can leave the app while it works, and Irdora sends a notification when the recording is ready. The notification gives the length and nothing that was said. If iOS stops the work before it finishes, it resumes when you open Irdora. A recording that never got saved because Irdora closed unexpectedly is added the next time Irdora opens, marked Recovered after Irdora closed unexpectedly.

Summaries and templates

A summary is shaped to the kind of conversation. Irdora suggests a template from the first minute, from 16 built-in templates in 4 groups: Health, Work, Learning and Personal. You can choose another at any time, and Settings has a Template for new recordings. You can also write your own template: a name, and up to 8 sections, each with a title and an optional one-line instruction.

Every summary opens with a title and In brief, a few sentences, then the sections of its template, then the action items. Add a summary in another template keeps the first and writes a second beside it. Rewrite the summary takes a short instruction, such as "shorter" or "focus on the medications". It can reorganise and condense what was said and adds nothing new. The version before is kept until you choose, and Undo the rewrite brings it back.

Figures are copied as said. Doses, dates, amounts and names come from the transcript, never from the model's own working, and a point that holds a number nobody said is discarded.

The transcript and speakers

The transcript shows who said what, with times. Tap a sentence to play the recording from that moment. Tap a point in the summary and the transcript opens where it was said.

Speakers are numbered at first. Tap a speaker's label to give a name. The name replaces the number everywhere in that recording: the summary, the mind map, the answers and the exports. Give 2 speakers the same name when they are one person. Who spoke shows how long each speaker talked, measured from the transcript.

Action items

Every recording lists its action items: what is to be done, who is to do it, and when, with the time it was said. The "when" is the phrase as said, such as "by Friday", and beside it is the date Irdora read from it, so a wrong reading is visible before anything is added. Check each date before you add it.

Add to Reminders adds a reminder, with the due date when there is one. Irdora asks for access to Reminders the first time, and afterwards reads only the reminders it added, to show which items are already there. Add to Calendar opens the system's own event editor with the event filled in, for you to check and save. An item shows when it was added, so it is not added twice.

The mind map and series

Each recording has a mind map of its topics and their key points. Tap a point to see where it was said. A timeline shows when each topic came up in the recording. The same content is available as an outline, which is what VoiceOver and the largest text sizes use.

A series gathers the recordings of a conversation that comes back: a weekly meeting, a course, a run of visits. When a new summary looks like a series you have, Irdora offers to add it, and nothing is added without your tap. You can also create a series with Create series and add recordings by hand. A recording belongs to at most 1 series. The series page lists its recordings by date and has Since last time: what is new, what is still open, and the topics that recur.

You can track a figure that was said, such as a blood pressure, and the series shows it on a chart by date. Nothing is tracked unless you ask, and nothing is worked out from the figures: no average and no trend.

Asking a recording

Ask this recording and Ask this series answer a question from what was said, on your iPhone. Every answer names where it came from, and tapping it opens the transcript there. If the answer is not in the recording, Irdora says That was not found in this recording instead of guessing. Asking is available in the languages that summaries support.

Listening, reading aloud and export

Play the recording from the player, or tap a sentence to hear it. On the Lock Screen, playback shows Irdora recording and the date, never the recording's title, because a Lock Screen can be read by anyone holding the phone.

Reading aloud has a summary or the transcript read to you in a voice installed on your iPhone. Choose the voice in Settings, under Voice, or leave it on Choose automatically.

To export, choose Export PDF or Export RTF from the recording's menu. You choose whether to include the action items, the mind map and the transcript, and Settings remembers the choice under Exports include. The summary and In brief are always included. The file is made on your iPhone and leaves it only when you share it. An RTF opens in Word, Pages and Google Docs.

iCloud sync

Sync with iCloud is off until you turn it on, when you first set up Irdora or later in Settings. When it is on, titles, transcripts, summaries, series, speaker names and your templates go to your own private iCloud database, so they reach your other devices. I cannot read that database. A recording's audio always stays on the device that recorded it, so on another device you can read, ask, export and have things read aloud, but not play the recording.

A change takes effect the next time Irdora opens. Turning sync off stops Irdora sending. What was already sent stays in your iCloud until you delete it in the iPhone's Settings, under iCloud storage.

Settings and purchases

Settings holds your subscription, with Restore purchases and Manage subscription, the notification status, your templates, the voice for reading aloud, what exports include, the iCloud switch, and the credits for the models Irdora runs on your device. If you start the free trial and Irdora's notifications are on, Irdora sends a notification the day before it ends.

What Irdora does not do

It does not record the audio of a phone or FaceTime call directly, because iOS does not allow it. It records through the microphone. It does not upload a recording's audio. It does not use a cloud AI service.

The on-device model is smaller than the models behind cloud services, so a summary can miss a point, and telling speakers apart is good and not perfect. The transcript is the record. For anything that matters, tap the sentence and hear it.