MemoryVerse AI '26

Student credentials — certificates, project docs, transcripts, portfolio links — pile up scattered across drives, emails, and folders over years. Nothing connects them, nothing surfaces the right one on demand, and manually organizing this is a chore students skip.

Chronicle ingests all of it — files and any pasted link — reads it natively (no OCR config needed), auto-categorizes it, and builds a relationship graph (a cert proves a skill, a skill is applied in a project, a project led to an internship). Retrieval is hybrid (intent-parsed filters + HNSW vector search + lexical boosting), so a plain-English or Hindi question returns a one-sentence answer with the original document one click away.

Telegram bot: Extends this to a chat interface— ask a question from your phone and get the answer back plus a button per matching record (original file, its links). Link-specific questions return the URLs themselves, each labelled by what they are (GitHub repo vs live demo vs credential), not just a filename. /links lists everything grouped by record; /link CODE re-links the chat to a different account. Linking uses a one-time 6-character code, and a chat maps to exactly one account — so once linked, ordinary words like "resume" are never mistaken for a stale code.