Installation

Prerequisites

Before you begin, make sure you have the following installed:

Tool	Version	Purpose
Rust	1.75+	Backend compilation
Node.js	20+	Frontend dev server
Supabase CLI	1.x	Local Supabase (PostgreSQL)
Docker	24+	Ollama and optional services
Ollama	latest	Local models and embeddings

Tip

If you only want to use cloud providers (OpenAI, Mistral, Gemini), Ollama is optional. However, the semantic context feature requires Ollama for generating embeddings with the nomic-embed-text model.

1. Clone the repository

git clone https://github.com/CITIUS-iRiA-eRisk/demo-inside-the-lab.git
cd demo-inside-the-lab

2. Start Supabase

The local Supabase instance provides PostgreSQL (with pgvector) and the Auth service.

supabase start

This command starts the database on port 54322 and the Auth API on port 54321. On first run, it automatically applies all migrations from supabase/migrations/.

After startup, note the output values — you will need:

• API URL (typically http://localhost:54321)
• anon key
• JWT secret

Note

If you need to reset the database, run supabase db reset. This drops all data and re-applies migrations.

3. Configure environment variables

Create a .env file in the backend/ directory:

backend/.env

DATABASE_URL=postgres://postgres:postgres@localhost:54322/postgres
SUPABASE_JWT_SECRET=<your-jwt-secret-from-supabase-start>
BASE_PATH=/chatty
OLLAMA_HOST=http://localhost:11434

# Add any API keys you have (all optional)
OPENAI_KEY=sk-...
MISTRAL_KEY=...
GEMINI_KEY=...

Create a .env file in the frontend/ directory:

frontend/.env

PUBLIC_BASE_PATH=/chatty
PUBLIC_SUPABASE_URL=http://localhost:54321
PUBLIC_SUPABASE_ANON_KEY=<your-anon-key-from-supabase-start>

See the Configuration reference for all available variables.

4. Pull an Ollama embedding model

If you want semantic context search, pull the embedding model:

ollama pull nomic-embed-text

You can also pull chat models to use through Ollama:

ollama pull llama3
ollama pull mistral

5. Start the backend

cd backend
cargo run

The backend starts on http://0.0.0.0:5000 by default. It automatically discovers any running Ollama models and registers configured cloud providers.

You should see output like:

INFO Starting Chatty backend on 0.0.0.0:5000
INFO Base path: /chatty
INFO Connected to database
INFO Registered Ollama models
INFO Total models registered: 3
INFO Listening on 0.0.0.0:5000

6. Start the frontend

In a separate terminal:

cd frontend
npm install
npm run dev

The frontend dev server starts on http://localhost:5173 by default (Vite).

7. Open the application

Navigate to http://localhost:5173/chatty in your browser. You will be redirected to the Google OAuth login page.

Caution

Google OAuth requires a configured OAuth client. For local development, see the Google OAuth setup section. Without it, you cannot log in.

Development workflow

After the initial setup, your typical workflow is:

1. supabase start (if not already running)
2. ollama serve (if not already running)
3. cd backend && cargo run
4. cd frontend && npm run dev

The frontend hot-reloads on changes. The backend requires a restart after Rust code changes (consider cargo-watch for auto-restart).