Privacy policy

Last updated 2 November 2026

Hey Atu lets you use the Pi agent on your own computer from your phone, or from an AI app such as Claude or ChatGPT. This policy tells you what we collect, why, and what we cannot see.

Who we are

Hey Atu is run by Xinyi Han, a sole trader in Australia ("we", "us"). Contact: privacy@heyatu.com.

The short version

What we collect

Website and waitlist

When you join the waitlist, we keep your email address, the page you used (home page or AI app page), a referral tag from the link if there is one, your country (from our host's network data) and the times you joined. We do not keep your IP address with this record. We use it only to tell you when early access opens.

You can join with a Google account. Google then tells us your email address and that it is verified. We do not keep your name, your picture or your Google account ID for the waitlist.

Service counts

To run and improve the service, the relay counts events: page views (the page, your country, the site that linked to us, a referral or campaign tag from the link, the type of device (phone, tablet or computer) and whether the visit looks automated), clicks on the main buttons of the website (for example "Join the waitlist"; never what you type), waitlist sign-ups, pairings, connections of phones and computers, AI app calls with their result code and duration, and refused or limited requests. A count has no content, no email address, no IP address, no device name and no browser details. A computer appears only as a one-way hash of its relay ID. We keep these counts for three months, and weekly totals for longer.

Request logs

Our host, Cloudflare, logs a sample of about 1 in 20 requests to the website and the relay, so that we can find faults and abuse. A logged request has its time, the page or route (without the part after "?"), the result, your IP address, approximate location (country and city) and browser type. It never has content, because the content is encrypted or is not logged. The logs are deleted after at most 7 days.

Usage statistics (you choose)

When you set up the agent, it asks if you want to share usage statistics. Yes is selected; you confirm or decline. If an AI agent sets up your computer for you, sharing is on unless it was told otherwise, and the result tells how to turn it off. With sharing on, your computer sends one report each day with counts only: messages sent, tasks finished, questions asked and answered (with the median time to answer), chats opened, features used, and the agent version and operating system. It never sends text, files, commands, folder names, model names or device names. The report has a random ID that is not your computer's relay ID; turning sharing off and on again makes a new one. See the exact report with heyatu metrics show. Turn sharing off with heyatu metrics off; your computer then deletes its counts at once.

Email to us

Mail to support@, privacy@ and security@heyatu.com goes to our mailbox. We keep it as long as we need it to answer you.

The phone app and the relay

To connect your phone and your computer, the relay keeps a random ID for each computer, a hash of each pairing key, and the name of each paired device (for example "iPhone"). It also handles connection data, such as when a device is online. It cannot read the content, because the content is encrypted on your devices.

The agent on your computer

The agent runs on your computer and keeps its data there. It downloads its updates from GitHub and Cloudflare. Your AI model provider (for example Anthropic or OpenAI) gets what Pi sends to the model. Their privacy policy applies to that, not ours.

iPhone data (you choose)

The iPhone app can read four kinds of data, each only after you turn it on and iOS asks for your permission:

The app only reads this data; it never changes your calendar, your reminders or your health data. It sends the data end-to-end encrypted through our relay to the computer that you paired, and that computer keeps it. We do not store it on our servers, and we cannot read it. Your agent uses it for the briefs and reports that you turn on, so your AI model provider gets the parts that a brief needs. We never use health data for advertising, and we do not sell or share it with anyone else. In the app, Settings → iPhone data → Delete removes the iPhone data of a source from your computer at once. Briefs, reports and chats that already used it stay on your computer until you delete them. You can also take back the permission in the iOS Settings app.

AI app connections (Claude, ChatGPT and others)

When you connect an AI app, you sign in with Google. We keep your Google account ID and email address, the names of your linked computers, and for each approved app: its name, its address, and when it was approved and last used. Access tokens are stored as hashes. The AI app's requests and the results from your computer pass through the relay in plain text. The relay forwards them and does not store or log them. The AI app's provider sees them; their policy applies.

Your account and linked computers

When you pair a phone with a computer, we make an account for that computer with no email and no name (an anonymous account), or we add the computer to the account of the phone's other computers. When you link a computer to Google (for Claude, ChatGPT or the account page, or with heyatu account link), its anonymous account joins your Google account. We keep the computer's relay ID, its name, when you linked it and when it last connected, in your account. An account has at most 3 computers. When you sign in to your account page, we set one cookie that keeps you signed in for 30 days. It is needed for the page to work; it is not used for tracking. We keep only a hash of it.

Payments

Stripe takes your payment on its own page. We never see or keep your card number. We keep billing records that Stripe gives us: a customer ID and a subscription ID, the plan status (for example trial, active or ended), the dates when the trial and the current period end, and whether your account already used its free trial, and whether it subscribed at the limited-time offer price. We also count the actions of an account without a plan (a number only), to give the free actions. Stripe gets your email address and a random account ID from us, not your Google account ID. If you pay in the iPhone app, Apple takes the payment under its own terms.

For the iPhone app, your account has a random ID (a UUID). The app gives it to Apple with a purchase, and Apple returns it with each transaction, so that we know which account paid. We do not get your Apple ID. From Apple we keep: the subscription's original transaction ID, the product, the plan status, the dates when the trial and the current period end, whether it renews, and whether it is a test (sandbox) purchase. The app reads your plan through your paired computer; it gets your account email (when the account has one), the plan and the number of free actions left. When you pair a phone, the app reads the current Hey Atu subscription of the Apple ID on that phone and sends it to us, so that your plan comes back on a new phone. The relay also counts the tool calls in your phone sessions (a number only; it cannot read them), with the AI app tool calls.

Email from us

We send email only to an account that has a Google address: a reminder 2 days before a free trial that you started in the iPhone app ends, and, from older versions of the app, a code when you ask to delete your account. An email service sends them for us. We keep a hash of the deletion code for 10 minutes. An anonymous account gets no email. The app can remind you of the trial end with a notification on your phone; the phone schedules it, and we send nothing for it. We send no marketing email.

Connecting an AI app from the menu bar

When you click Claude or ChatGPT in the Hey Atu menu bar, we keep for 10 minutes which app you chose and a one-way hash of your network address. The next sign-in of that app on your account then needs no approval. If someone signs in with another Google account on the same network in those 10 minutes, the page shows the email of this computer's account in part only (for example k•••@gmail.com).

What we do not collect

We do not collect the content of your chats, files or commands. We do not use cookies for tracking, analytics or ads; the account page uses one cookie to keep you signed in. Our team's own browsers carry one more cookie, so that we do not count our own visits; we do not set it for visitors. Our own script on the website counts button clicks; it sets no cookie and keeps no ID. The website loads one script from another company: the Google sign-in button. It loads only when you open the waitlist sheet. Google may process data from your browser when it loads, under its own privacy policy. The website loads no other scripts from other companies. We do not use another company's analytics service; counts stay in our Cloudflare account.

Who helps us

These companies process data for us under their own terms. Cloudflare and Stripe run global networks, so your data can be processed outside Australia.

How long we keep it

Your choices

If you are not satisfied with our answer, you can complain to the Office of the Australian Information Commissioner (oaic.gov.au) or to the privacy authority where you live.

Children

Hey Atu is not for people under 18. We do not knowingly collect data from them.

Changes

We will update this page when our practices change, and change the date at the top. For a material change, we will tell people on the waitlist or in the app before it applies.