@DoublegramScribeBot builds a searchable memory of each Telegram chat where it is active. You do not export logs or run commands — indexing happens automatically in the background while members talk.
This guide explains what the bot remembers in Telegram groups (and the same logic applies to your private chat with @DoublegramScribeBot).
What gets remembered
While Scribe is in a group, it records messages and events it can see:
- Text messages — Saved as soon as they arrive
- Voice notes — Saved after transcription (the spoken text, not just the audio file)
- Photo and video captions — Text you attach to media
- Replies and forwards — Who replied to whom, when relevant metadata is available
- Member events — Joins, leaves, and admin changes (as structured events)
- Forum topics — Messages stay linked to their topic thread
Each entry includes who sent it, when, and what type of message it was — so Scribe can filter by person or date later.
Scope: Scribe only remembers messages in chats where the bot is present. It cannot see other groups or private chats it was never added to.
How people are recognized
When someone writes in the group, Scribe learns their Telegram username and display name. That lets you ask later using natural names — What did Marco say? or Messages from @nebula — without memorizing user IDs.
The index builds from real activity: someone must have posted in the group for Scribe to recognize their name reliably.
Two layers of memory
Scribe uses complementary storage so search feels instant:
- Recent cache — The latest messages (up to about 100, kept for several hours) are available immediately for search
- Long-term archive — Older messages are stored for semantic search across the full history of the chat
New messages appear in the recent cache right away. They move into the long-term archive after a short processing window (usually seconds to a couple of minutes). Scribe merges both when answering — so what did they say just now? still works.
What Scribe does not store for search
- Image or video content — Only captions and transcriptions; Scribe does not analyze pixels for search
- Audio file attachments — Use voice messages (microphone button) so Scribe can transcribe and index speech
- Messages before the bot was added — History starts when Scribe joins
- Other chats — Each group and private chat is isolated
Using memory in groups
Transcription is free in groups. Searching history requires group premium (AI feature): mention @DoublegramScribeBot with your question. Without premium, Scribe transcribes voice but does not answer history questions.
In private chat, history search needs a personal subscription — just send your question directly to the bot.
Quick reference
- Automatic: no manual export
- Voice → indexed as transcribed text
- Per-chat only — no cross-group access
- Group search: mention bot + group premium
- Private search: message the bot directly
Related: example questions you can ask about chat history in Asking Questions About Chat History in Telegram.