Analyze a chat without giving it away.

Local-first means the file stays here.

A browser can read and calculate over a file without uploading it. In Telescope’s local analysis, the export is parsed in the page’s memory on your device. Counts and charts are created there; the raw conversation does not need to cross the network.

Saving a result is different from uploading a source.

When you sign in, Telescope can save the completed numerical analysis so you can return to it. That result contains aggregates and report data; it is not the original Telegram export.

Shared links are another separate choice. Treat every public link as something its recipient could forward, and review whether message evidence is included before sending it.

AI reading is an explicit upload.

A model cannot interpret the writing without receiving conversation content. Telescope keeps that step separate from local analysis and labels it before it happens. If you do not request AI insights, the numerical report remains usable without that upload.

This distinction is worth checking in any chat-analysis product: “private” can describe account access while the raw file is still processed on a server. Look for a precise explanation of where parsing occurs and which actions initiate network requests.

Before choosing a conversation.

1. 01Use a device you trust and remove old exports when you no longer need them.
2. 02Check whether the tool processes the raw file locally or uploads it by default.
3. 03Treat AI analysis as data sharing and decide whether the conversation is appropriate to send.
4. 04Review message visibility before creating or forwarding a shared report.
5. 05Remember that the other person did not write their messages for an analytics system.