Hopper · Legal
Privacy Policy
Hopper is an AI teammate a company adds to its Slack, GitHub, Linear and calls. To be useful it reads what the company lets it read, so this page is specific about what that is, where it goes, how long it stays, and what you can do about it.
In effect since 25 October 2026 · version 2026-10-25
01Who is responsible
For your account, and for anything you send us directly, the controller of your personal data is Kocsmarszky Zsolt, an individual established in Hungary.
For what a company's Hopper reads at work (its messages, calls, issues and code), the company that set Hopper up decides what it may reach and why. That company is the controller, and Hopper processes that content on its behalf, as its processor. If you are on such a team, your company is the first place to ask; we will help it answer you. A company can ask for a data processing agreement at the address below.
Contact: Kocsmarszky Zsolt, 1039 Budapest, Lukács György utca 18. 5/16, Hungary · hello@poolplanner.app
02Your account
When you sign up and set up a Hopper, the following is stored:
- Your name and email address, and your password, kept only as a scrypt hash that cannot be turned back into it. If you continue with Google, the name and address Google confirms take the place of a password. To create the account and sign you in. Legal basis: performance of a contract.
- Whether your address is confirmed. Confirmation links work for 24 hours and password-reset links for one hour. Each works once and is stored only as a SHA-256 hash. Legal basis: contract.
- Your sessions. A signed-in browser stays signed in for 30 days, and the browser's user agent is kept with the session so it can be recognised. Legal basis: contract, and legitimate interest in keeping accounts secure.
- Your company: its name and website, what you told setup about the team, and your Hopper's settings: its name, persona, personality and autonomy level. Legal basis: contract.
- Invitations: the address of someone you invite, and who invited them, kept for seven days or until the invitation is accepted or withdrawn. Legal basis: the inviting company's legitimate interest in adding its team.
- Your IP address is used for a few minutes, in memory, to slow down repeated sign-in and password-reset attempts. It is not written to our logs or to the database. Legal basis: legitimate interest in security.
03What your company's Hopper reads
All of this happens only for the tools your company connects, and only inside them.
- Slack: messages in the channels Hopper has been added to, and in its direct and group messages, with who wrote them (their Slack name and email address, so it can tell teammates apart). The text of Slack messages is deleted after 14 days. What Hopper learned from them stays in its memory.
- Calls: Google Meet calls Hopper is invited to or sent a link to. It introduces itself when it joins. Its transcript is Google Meet's own live captions, with speaker names, and after the call it writes a recap. Invitations on its calendar carry the meeting's title and time and the addresses of its organiser and guests.
-
GitHub: the repositories your company chose when installing Hopper's GitHub app, and the issues, pull requests, comments and
reviews it works on. For each coding task the code is checked out into an isolated sandbox that holds no credentials, and the sandbox
is deleted when the task ends. Hopper writes only to its own
hopper/branches and never merges. - Linear: issues and comments in the connected Linear workspace: those from the last 30 days when Linear is connected, then each change.
- Your company's public website, read during setup to draft a description of what the team builds.
- Memory: what Hopper keeps from all of the above (summaries, decisions, action items with their owners, and a short team brief), so it can answer later and point to where something was said.
- Model call records: what was sent to the AI model and what came back, with token counts and cost, for debugging and billing. Kept for 90 days.
Legal basis: the company's instructions, under its contract with us. The company is responsible for having its own legal basis, and for telling its team and the people in its calls that Hopper is there.
04AI models and training
Hopper's reading and writing is done by AI models from Anthropic. When Hopper speaks in a call, OpenAI turns its words into its voice; OpenAI receives only what Hopper says, not what anyone else in the call says.
Nothing you or your team share is used to train AI models: not by us, and not by Anthropic or OpenAI, whose terms for API customers exclude training on what they receive.
05Who else processes it
These providers process data on our behalf, under contract:
- Hetzner Online GmbH
- Hosting. Every database, file and log is stored on its servers in Nuremberg, Germany.
- Anthropic, PBC
- The AI models that read and write for Hopper (United States).
- OpenAI
- Hopper's voice in calls. It receives only Hopper's own words (United States).
- Resend
- Account email: address confirmations, password resets and invitations (sent from its EU region in Ireland; Resend is a US company).
- "Continue with Google", if you choose it, and the Google Meet and Calendar of Hopper's own meeting account.
- Paddle.com Market Ltd
- Our reseller and merchant of record for payments (United Kingdom). It handles card details; we never see them.
Slack, GitHub, Linear and Google Meet are your company's own tools, used under its own agreements with them. Hopper reads from them and writes to them only as your company directs.
Data is stored in the European Union. It leaves the EU only for the US providers above, under their standard contractual clauses, and for Paddle in the United Kingdom, which the EU recognises as giving adequate protection.
07Security
Everything travels over TLS. Passwords are stored as scrypt hashes. The Slack and Linear access tokens Hopper holds are sealed with an encryption key before they are stored. Coding runs in sandboxes isolated from the server and from each other, with no credentials inside and no way to reach our private network.
08How long it is kept
- Your account: until you ask us to delete it, and then it is gone within 30 days.
- What your company's Hopper read and remembers: for as long as the company uses Hopper, and deleted within 30 days after the company asks or closes its account.
- The text of Slack messages: 14 days. Model call records: 90 days. Coding sandboxes: until the task ends.
- Confirmation links: 24 hours. Password-reset links: one hour. Invitations: seven days.
09Your rights
Under the GDPR you may ask for a copy of your data, ask for it to be corrected or deleted, object to or restrict processing, withdraw consent where processing rests on it, and ask for your data in a portable form. Write to the address above and we will answer within a month. If your data reached us through your company's Hopper, we may pass your request to the company, which decides about its workspace.
You may also complain to a supervisory authority. In Hungary that is the Nemzeti Adatvédelmi és Információszabadság Hatóság (NAIH), Budapest.
10If you are in the United States
We do not sell personal information and do not share it for cross-context behavioural advertising; there is no advertising here at all.
California residents may request to know what personal information is held, request its deletion or correction, and are entitled not to be discriminated against for exercising those rights. Residents of other US states with comparable laws have comparable rights. Use the contact address above; we will verify the request against the email address on the account.
11Children
Hopper is a tool for businesses. It is not directed at children, and accounts are not knowingly created for anyone under 16. If you believe a child has an account here, write to us and it will be removed.
12Changes
This policy was last changed on 25 October 2026. If it changes substantively it will be dated again here, and the owners of companies on Hopper will be told by email.
Questions about this page go to Kocsmarszky Zsolt, hello@poolplanner.app.