Ask for your campus schedule.

Log in with your student username and password — the bot looks you up and answers about your lecture or practical days, campus, time, subject and room.

Try it now in-browser demo — the same dialogue and data as the deployed bot, replayed without a server

Username: first name + last name · password: last name + first name, run together — e.g. alicejohnson / johnsonalice. Case and spaces are ignored. Checked in your browser against the same 15 seeded students; the live deployment validates against its database.

How the real bot works

Rasa 3.6 — NLU & dialogue

A DIET classifier extracts names and intents from natural sentences ("I'm Bob Smith", "good morning, Alice Johnson here"). Rule, memoization and TED policies drive the conversation — 12 intents, 19 stories, a trained model checked into the repo.

Custom action server

Eight custom actions (rasa-sdk) answer schedule questions. They are stateless — they read the student's name from dialogue slots and query the attendance database on demand, so every answer reflects the current data.

One database, two backends

Locally the data lives in SQLite. In a hosted deployment the same queries run against Neon Postgres — the app switches when DATABASE_URL is set, with no query changes.

The real deployment

github.com/oldspoon/Rasa-AI-Assistant — the full monorepo: Rasa project, action server, React web chat, deployment files and tests.
Web chat on GitHub Pages — the React UI calling the live Rasa REST channel (/webhooks/rest/webhook).
Rasa server on Hugging Face — a Docker Space running Rasa + the action server together, trained during the image build. Currently parked on the free plan: Docker Spaces need compute access, which requires a paid plan.
Neon Postgres — the hosted attendance database (15 seeded students). The Space connects to it through the DATABASE_URL secret.

Run it locally

Python 3.10 required (Rasa 3.6.x). Full instructions in the repository README.

git clone https://github.com/oldspoon/Rasa-AI-Assistant
cd Rasa-AI-Assistant
py -3.10 -m venv venv
venv/Scripts/python.exe -m pip install -r requirements.txt
venv/Scripts/python.exe action_server/database/populate_db.py

# Terminal 1 - the action server (leave running)
cd action_server && ./dev-run.sh -r

# Terminal 2 - chat with the bot
cd rasa && ./dev-run.sh -r