the privacy of your data is important to us. we will not sell your data or use it in nefarious or abtrusive ways.
your message data is not kept in our database once it is synced to your graph
we keep your messages secure until you open roam, fetch them, put them in your graph, then delete them from our database. monthly backups are an exception to this. if this feature is enabled, we keep messages around long enough to generate the backup, then delete the data at the end of the month. we keep meta data of messages around for our metrics and diagnostics, but the text of your messages is not saved anywhere after it gets to roam. we receive secure URLs for attachments such as images and audio from twilio, which we also delete after syncing. we are keeping a giant list of URLs sent to us because having this aggregate data is useful and interesting, but we dissassociate links from users after sync.
we might want to keep more message and attachment data around to do fun machine learning stuff like OCR and NER, but these will be optional, off by default, and more information will be added to this page to help people learn more before opting in.
clerk handles most of our user authentication, and verifying phone numbers and emails. we store only a clerk_id and ptn_key for each user into a secure postgres database.
when you send us a message with a URL, our system fetches the contents of the URL, and parses metadata in the response using OpenAPI standards. the meta data of the link is stored in our database.
please contact us if you have any questions or would like any data deleted.
minimal extra data is collected for the helpful keyboard iOS app. here's how autocomplete works:
- we won't do anything until you turn on page indexing from your `roam/js/ptr` configuration in your roam graph
- once that's turned on, we'll index your page titles, and hold them in our private database
- when you use our keyboard, we use your roam key to securely fetch those page titles for autocomplete
- we use what you're currently typing (just the current word) to search your indexed page titles
we send minimal data back to phonetonote for analytics and bug logging.