Your words stay yours.

Who we are

Capture is made by Graviti (“we”, “us”). This policy covers the Capture app for iPhone, the Capture server it talks to, connecting AI apps to Capture over MCP, and this website. We are responsible for your information as this page describes.

Questions and requests go to help@graviti.inc.

What we collect

Your account

When you sign in with Apple or Google, we receive your name, your email address (your Hide My Email address, if you chose one) and an identifier Apple or Google gives us for you. We also keep your time zone and your settings.

Your recordings

Capture records only while a capture you started is running. While you speak, the audio streams to our server so it can be transcribed as you talk; when you stop, the recording is uploaded. The settings it’s processed with (language, task style, default category and snippets) are stored with it.

What Capture makes from them

Transcripts, tasks and their checklists, due dates, repeats, priorities, reminders and alarms, your categories and their descriptions, and your Dictionary: the words you’ve taught Capture, and the words it suggests along with the bit of transcript each suggestion came from. We also keep daily counts (recordings, words, tasks) for your Stats screen, and a record of what each AI call cost us.

About your work, if you choose to add it

“Make Capture yours” asks where you work and what you do, and, if you like, for your LinkedIn address. With a LinkedIn address, your phone opens that public profile page itself and sends us the address along with the page’s title, summary and profile data. Our server reduces the page to a few lines (headline, employer, job titles, schools, location, About, recent post titles), removing your name, links, email addresses and past employers. The AI gets those lines and the address, which usually contains your name, so that it can look the profile up.

We use all of this once, to suggest categories and Dictionary words, and we don’t store it on our server. Your phone remembers the company and role you typed, but not the LinkedIn address.

Your devices

For each phone signed in to your account, we keep an identifier for that installation, what kind of device it is and when it last checked in. Capture doesn’t send push notifications yet. When it does, we will also keep the phone’s push token, and the notifications will carry none of your content.

Server logs

Each request to our server is logged with your IP address, your account identifier, a request identifier, the address requested, the time and the result. Some events add a detail, such as the domain of a waitlist address or your time zone. Logs are written to leave out your recordings, transcripts and tasks.

Crash reports

Crash, hang and launch reports from iOS stay on your phone. They leave it only if you export them and send them to us yourself.

Email to us

If you write to us, we get your message and anything you attach. Emails the app starts for you add the app’s version and your iOS version at the bottom.

The waitlist

If you leave your email address on this site, we keep the address, which form it came from and when. We use it to send you one email when Capture ships.

This website

capture.graviti.inc sets no cookies and loads nothing from anyone else. Cloudflare, which hosts it, sees your IP address in order to serve the page.

What we don’t collect

Capture doesn’t ask for your location, contacts, photos, calendar or camera, and it doesn’t use Apple’s speech recognition. It has no advertising identifier. There is no advertising, analytics or tracking code, and no third-party crash reporting, in the app, on the server or on this site. Nothing tracks you across other apps or websites.

How we use it

• To do what Capture does: transcribe what you say, turn it into tasks, file them, remind you, and keep your phone and our server in step.
• To keep Capture working and safe: finding and fixing faults, and preventing abuse.
• To answer you when you write to us.
• To send the one launch email, if you joined the waitlist.

We don’t sell your information, we don’t use it for advertising, and we don’t use your recordings, transcripts or tasks to train AI models. Nobody at Graviti reads your recordings, transcripts or tasks except when you ask us to (for example, to look into a problem you reported), when it’s needed to keep Capture working (for example, to process again a recording that failed, and then only as much as needed), or when the law requires it.

If you are in the EU, the UK or somewhere with similar law: running Capture for you is how we carry out our agreement with you. “Make Capture yours” runs on your consent, which you can withdraw by not using it. Logs and fault-finding rest on our legitimate interest in a reliable, secure service. And we keep what the law requires us to keep.

How the AI works

Transcription and task-making run on Amazon Bedrock, using Amazon’s Nova models, in AWS’s United States regions. For each recording we send the audio and its transcript. So the tasks land in the right place, we also send your Dictionary, your category names and descriptions, your task style, your snippets, and the time and time zone of the recording, so that “tomorrow at 9” lands on the right day.

For “Make Capture yours” we send the company and role you typed and, if you gave a LinkedIn address, the address and the lines described above. The model may search the public web for your company and for that profile to find the words it uses.

Under Amazon’s terms, Bedrock doesn’t store what we send it or what comes back, and Amazon’s staff have no access to it. If we ever change the AI provider, we will update this page before any of your information goes to the new one.

AI apps you connect

Capture is an MCP server. When you connect ChatGPT, Claude or another app that speaks MCP, you choose read-only or read & write. Either way, that app can see your tasks, categories, transcripts and recordings. With read & write it can also change your tasks and categories, and add words to your Dictionary. See every tool.

What the app then does with your information is up to its own privacy policy, not this one. You can disconnect any app in Settings › Connections › MCP. Deleting your account disconnects them all; signing out does not.

Connecting an app shows a sign-in page from our server. That page loads Apple’s and Google’s sign-in code and a font from Google, and it sets one cookie, used only to finish signing in.

Where it’s kept, and how

Your account, transcripts and tasks are held in a database, and your recordings in private file storage, both with Amazon Web Services in Mumbai, India. The AI runs in the United States, as described above. So your information is processed in India and the United States, whose data-protection laws may differ from yours. Where the law requires safeguards for this, we rely on our providers’ standard data-protection terms.

• Everything between your phone and our services travels encrypted.
• Recordings are encrypted in storage, and the storage is not public.
• Your sign-in stays in your phone’s Keychain; our server keeps only a scrambled form of it that can’t be used to sign in.
• Your phone keeps its own copy of your tasks so that Capture works offline. It keeps each recording for 7 days after the recording has uploaded, or for up to 30 days if it couldn’t be uploaded, and leaves recordings out of device backups.

No system is perfectly secure. If a breach affects your information, we will tell you as the law requires.

How long we keep it

• Recordings. By default, until you delete them, so that you can play them back. Settings › Data and privacy › Keep recordings deletes the audio after 7, 30 or 90 days, or within a day of transcription. Transcripts and tasks stay either way.
• Delete all recordings, in the same place, deletes every recording and its transcript from our server and your phone. Your tasks stay. A recording still being processed at that moment is left, so delete it again once it has finished.
• Transcripts, tasks, categories and Dictionary. These stay until you delete them or your account. When you delete a task, category or Dictionary word, we keep it hidden for 90 days so that your other devices can drop it too, then remove it for good. A deleted recording loses its transcript at once; an earlier draft of it, kept when a recording is transcribed again, goes at the 90-day mark. A Dictionary suggestion you decline is kept, with the few words it was heard among, so that it isn’t suggested again. It goes when you delete your account.
• Your account. Deleting it removes everything above from our server at once, and wipes Capture’s data from the phone you delete it on.
• Backups. A deleted recording can remain in our storage’s version history for up to 30 days. Database backups are kept for 7 days, and we take a snapshot before each server update, keeping the latest five. Deleted information disappears from those as they are replaced. If we ever restore a backup, we will delete again anything you deleted since it was made.
• Server logs. These are kept by our host for a limited time, then discarded.
• The waitlist. We keep your address until you ask us to remove it.
• Email to us. We keep it for as long as we need it to help you.

Who else handles it

Amazon Web Services

Hosts our server, database and recordings (India) and runs the AI (United States).

Apple

Sign in with Apple. Reminders and alarms are scheduled by iOS on your phone.

Google

Sign in with Google, if you use it. Google also hosts the inbox our email arrives in, and serves the font on the page for connecting AI apps.

Cloudflare

Hosts this website, and forwards email sent to help@graviti.inc.

LinkedIn

Only if you give Capture your LinkedIn address. Your phone fetches the public page directly, so LinkedIn sees your phone’s IP address, as it would for any visit.

AI apps you connect

Only the ones you choose, as described above.

Otherwise we share your information only when the law requires it, when it’s needed to protect someone’s safety or our rights, or as part of a merger or sale of Graviti. In the last case, this policy goes on applying to your information.

Your choices and rights

In the app:

• See and correct. Your tasks, categories, Dictionary and recordings are all in the app, and you can edit or delete them there.
• Export. Open the menu, then Account › Export my data. This makes a file of your account, categories, tasks, Dictionary and transcripts, plus any crash reports your phone has kept. For your recordings, or anything the file leaves out, email us.
• Delete. Open the menu, then Account › Delete account. It takes effect at once and can’t be undone.
• Limit. Set Keep recordings, skip “Make Capture yours”, or disconnect AI apps.

Depending on where you live, for example in the EU or UK, India or California, you may have the right to:

• access, correct, delete or export your information
• restrict or object to how we use it
• withdraw your consent
• complain to your data-protection authority

Email help@graviti.inc and we will answer within 30 days. Asking won’t change how we treat you. We don’t sell or share personal information as California law defines those terms.

Children

Capture isn’t meant for children under 13, or under the minimum age your country sets. If you think a child has given us information, tell us and we will delete it.

Changes to this policy

When this policy changes, we will update the date at the top. If a change affects how your information is used, we will tell you in the app before the change takes effect.

Contact

Graviti, help@graviti.inc