Your street is nobody’s business.

Paper Trail has no accounts, no analytics, no advertising and no tracking cookies. It keeps nothing about you on its server. This page explains exactly where your information goes when you use it.

What stays in your browser

Your trails, the reports from your photos, your Trail Book and your settings are stored in this browser (IndexedDB and local storage). They never leave it unless you send them somewhere yourself. Clearing your browser’s site data deletes them, and nobody, including us, can recover them.

What passes through the server, and where it goes

• Where you start a walk. The coordinates you pick are sent to OpenStreetMap services (Overpass for nearby places, Nominatim if you search an address), to OpenFreeMap for the same OpenStreetMap streets and places as map tiles (a backup when Overpass is slow; it only sees which ~2 km square you are in), and to Open-Meteo for the weather. Results are cached on the server for up to a day, keyed by rounded coordinates, so that repeat lookups don’t hit those services again.
• What you type in the search box. As you type (from the third letter), the text is sent through the server to Photon, komoot’s OpenStreetMap search, for suggestions. So that nearby places come first, it also passes a rough point, rounded to about a kilometre: the spot you picked, your last trail’s start, your position if you have already allowed location, or else just the region of your time zone. Your IP address isn’t passed on.
• Writing your zine. A list of the nearby places, the weather, the walk length, who’s coming and the name for the cover (if you give one) are sent to the AI model that writes the pages.
• Reading your sheet. Your browser crops your photo into four page images on your device; only those four crops are sent to the AI model that reads them. The full photo is never uploaded.
• The map picker. If you open the optional map, its tiles load directly from openstreetmap.org.

Which AI model, and who runs it

Paper Trail always uses Gemma, an open model. On this site it is served by a hosted provider, so the requests above go to:

• Google (Gemini API)
• Ollama (Ollama Cloud)

Those requests are handled under the provider’s own terms, which may allow it to keep them for a time or use them to improve its services, depending on the plan in use. If you would rather nothing leave your computer, you can run Paper Trail yourself with Gemma on your own machine.

Logs

For each request the server writes one technical log line: which page or API was used, whether it worked, how long it took and a random request number. These lines never contain your location, places, photos, handwriting or report. If a page crashes in your browser, it reports the error message and the page address (without any query string) so we can fix it. The hosting provider (Render) also keeps standard access logs, which include IP addresses, for a limited time; we use the IP address only in memory to stop any one visitor from overloading the service.

Children

Paper Trail is made for families to use together. It has no accounts and collects no personal information. If a child uses it, a grown-up should choose the starting point.

Questions

Ask anything on the project’s issue tracker, or read the code: everything described here is in the public repository. See also the terms.