Yes. Open @Emcbearbot on Telegram and accept the Terms to start — no invite code required. You must be 18+ (or use Emily with a parent or guardian who accepts the Terms on your behalf). We're in **pilot mode** — expect active tuning.

No. Message Emily like you'd text a friend — plain sentences work for studying, directions, reminders, and memory. Shortcuts like /help exist if you want them.

Yes. Send Telegram voice notes or lecture recordings — Emily transcribes them for tutoring and proofreading. After a lecture summary, say save this lecture to my memory to keep key points (opt-in). Replies are text by default; optional voice replies follow when voice is on. Say voice off or text only for faster text-only replies.

Yes — preferences, professors, learning goals, and lecture notes you choose to save live in a readable memory tree. Emily also scouts what she already knows about you before each reply. You stay in control: privacy mode on, what do you know about me?, export my data, or forget everything.

Only when you share it in Telegram (attach → Location). Emily keeps it briefly (~15 minutes) for directions, weather, arrival reminders, and Uber estimates — not continuous tracking. Family check-ins are separate and student-initiated.

Yes. Like any AI, Emily can be incomplete or wrong — especially during the pilot. Campus and file-based answers use your school's information and materials you share; Emily may say when she is uncertain. If you get a self-correcting message, wait about 10 minutes and try again — she may hold a bad answer and try an automatic fix, or flag it for a human. Use Emily to learn and organize — not as the final word on grades, medical, or legal matters.

Emily is live for early users at Western's science program. You may see rough edges and slower response times while we tune routing, quizzes, and campus answers. Replies may take longer than usual while we run extra accuracy checks and tune routing — especially during busy periods. During the pilot, Emily checks replies before they reach you. If she catches a logic issue on her side, she may hold that answer and try an automatic fix — or flag it for a human. If you get a self-correcting message, wait about 10 minutes and try the same question again. Not every issue fixes itself. If it still fails after a retry, say so in chat or email support.

During the pilot, Emily runs extra accuracy checks before sending answers and may pause when she detects a logic issue. That refinement work can make replies slower than a typical chatbot — especially at peak times. If you see a self-correcting message, wait about 10 minutes and try again; not every issue fixes itself.

No. Emily is built for university students with study modes, campus help, memory you control, published policies, and clear boundaries — not a general chat app.

Allowed when genuinely academic; blocked when sexualized or romantic roleplay.

Safety filtering yes; surveillance no. Violations logged locally 30 days. Trusted contact only for crisis or spend alerts.

Refused, logged locally, no parent alert unless mental-health crisis.

Say false alarm or send /appeal in Telegram. Emily logs the appeal for parent review via the rescue pipeline — not an instant bypass of real safety boundaries.

Say "I have feedback: …" or /feedback in Telegram to reach the Emily team. For beta bugs you can also email [email protected].

Yes — say "forget everything", use private mode, or ask to delete specific items. A full wipe may require a forget PIN if a parent enabled one. See Privacy Policy for exceptions (e.g. files already in your Google Drive).

No — augmentation for after-hours clarification and practice.

Roadmap. Contact [email protected] for pilot interest.

Checklists, deadlines, essay guidance — not ghostwritten essays. Use proofreading on your draft.

Research and official links — not resume/cover-letter writing.

Opt-in student-initiated check-ins — not continuous tracking.

Opt-in, encrypted, private scope by default.

Assignments, classes, Drive links, reminders — when OAuth configured.

Primarily on dedicated infrastructure you connect to via Telegram — local-first JSON and SQLite, not a multi-tenant student cloud. Optional copies go to your Google Drive when you export. See Privacy Policy.

Yes. Say "export my data" in Telegram for a ZIP (about once per 24 hours) with chats, memory, quiz progress, and more. You can also save chats to Google Drive — see /trust/safety/#backups.

Partially. When Google Drive is connected, Emily auto-syncs your chat transcript, latest session, notes, and timetable to your Drive exports/ folder after you chat. Memory and quiz progress do not auto-sync — use full export. Emily does not guarantee data can never be lost; export copies you keep are your safety net.

Saying "forget everything" wipes data on Emily's server, but files already in your Google Drive stay there. Export first if you want a complete archive before forgetting.

Emily uses AI services to generate replies. By default, your name and personal details are replaced with generic labels on Emily's server before those calls — the real mapping never leaves Emily, and replies are checked so labels do not leak back to you. See /trust/safety/#privacy-shield. Subprocessors are listed at /subprocessors.

If an AI provider only saw partial redactions (like J*** S***), someone could piece clues together across many queries. Emily uses fixed generic labels instead — and keeps the real name map on her server only — to reduce that risk. On by default.

Yes. Before chat (except /help and /emergency), you confirm you are 18+ and agree to the Terms and Privacy Policy.

Emily ships with campus knowledge for your configured institution. Pilot default: **Western University in London, Ontario** — especially the **science program**. Ask about Weldon Library, science buildings, or transit; generic university answers are a fallback.