What Data Does Raven Actually See When You Upload a Photo?

Vague reassurances like "we take your privacy seriously" don't actually tell you anything. So instead of a general promise, here's a specific, step-by-step account of what technically happens to an image file from the moment you select it on your device to the moment Raven's server has finished with it — because being specific is the only way transparency actually means something.

Step One: What Leaves Your Device

When you choose a photo to upload, your browser reads that file as raw bytes — the same underlying data as the file sitting on your phone or laptop, nothing added or extracted from it locally. That data travels to Raven's server over an encrypted HTTPS connection, the same kind of connection used for online banking or any other site that handles anything sensitive. Nothing about the upload step is unusual or specific to Raven; it's the standard way any web application receives a file.

Step Two: What the Server Actually Checks

Once the image arrives, the server doesn't just trust the label your browser attached to it. It checks the file size against a limit, and it inspects the actual bytes of the file — its real internal signature — to confirm it's genuinely an image type the analysis can handle, rather than trusting whatever content-type your browser claims it is. This matters because a mismatched or spoofed file type is a classic way to smuggle something unexpected past a naive upload check, and it's worth doing properly rather than skipping.

Step Three: Where the Image Actually Lives (and Doesn't)

Here's the part that matters most: the image is held only in your request's active memory. It gets converted to a text-safe format in that same memory and handed directly to Google's Gemini model for analysis. There is no intermediate step where the file touches a hard drive, gets written to a cloud storage bucket, or gets inserted as a row in a database. The moment the request finishes — success or failure — that memory is released, and the image data goes with it. If you upload the same photo twice, the second request starts from zero; there is no stored copy anywhere to compare against or retrieve.

What About Location Data Hidden in the File Itself?

Photos can carry embedded metadata — EXIF data — including precise GPS coordinates, camera model, and timestamp. Raven's analysis doesn't read or rely on that metadata at all; the guess comes purely from what Gemini can see in the pixels, the same way a person looking at the photo would reason it out visually. If you're ever uploading a photo anywhere and you're specifically worried about embedded GPS data riding along in the file, it's a reasonable habit to strip that metadata first using your phone's share settings or a dedicated tool — not because Raven stores or exposes it, but because it's good practice generally, especially for services you're less familiar with.

What Actually Gets Logged

Some things genuinely persist, and it's worth being clear about what they are so there's no confusion. Your account information — email and basic profile details from sign-in — is stored, because that's how your account and free-trial status work at all. A record that you've used your free guess gets saved, again because that's the whole point of a free trial. Ordinary operational logs — timestamps, error messages, response times — help catch bugs and keep the service running, the same way almost every web service logs basic request metadata. None of that includes the image itself or any of its pixel content, at any point, in any log.

Why Bother Being This Specific

This is the same underlying philosophy behind Geospy AI, Raven's sibling app on the App Store — visual analysis that reasons from a photo in the moment rather than building a retained archive of one. Neither product needs to remember your photo to answer the one question it's built to answer: where does this look like it was taken? Once that question has been answered, there's nothing left worth keeping, so nothing gets kept.