Local Setup

1. Clone the Repository

git clone git@github.com:Monsoft-Solutions/loquent-app.git
cd loquent-app

2. Configure Environment

Loquent uses two separate env files:

File	Purpose	When read
.env	Runtime config (database URL, host)	Every app start
seed.env	API keys and credentials for initial DB seed	Once, during first migration

cp .env.example .env
cp seed.env.example seed.env

.env — Runtime

APP_HOST=localhost:8080
DATABASE_URL=postgres://user:password@localhost:5432/loquent

Variable	Description
APP_HOST	Bind address for the Axum server
DATABASE_URL	PostgreSQL connection string

seed.env — Migration Seed

Fill in all fields. The migration will abort if any value is empty.

TWILIO_MAIN_SID=your_twilio_account_sid
TWILIO_MAIN_TOKEN=your_twilio_auth_token
OPENAI_API_KEY=your_openai_api_key
RESEND_API_KEY=your_resend_api_key
EMAIL_DOMAIN=your-domain.com
EMAIL_USER_ADMIN=admin
GOOGLE_PROJECT_ID=your-gcp-project
GOOGLE_CLIENT_EMAIL=your-service-account@project.iam.gserviceaccount.com
GOOGLE_PRIVATE_KEY="-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n"

Caution

GOOGLE_PRIVATE_KEY must be wrapped in double quotes. PEM keys contain newlines that break env file parsing otherwise.

After migration, these values live in the core_conf database table. seed.env is never read again at runtime.

3. Create the Database

The migration tool connects to an existing database — it won’t create one for you.

createdb loquent

Or via Docker:

docker run -d --name loquent-db \
  -e POSTGRES_DB=loquent \
  -e POSTGRES_USER=user \
  -e POSTGRES_PASSWORD=password \
  -p 5432:5432 postgres:16

4. Run Migrations

Creates all tables (51 migrations) and seeds the core_conf table from seed.env:

just migrate

This runs sea-orm-cli migrate up. Key tables created:

Table	Purpose
user, session, member	Auth and identity
organization	Multi-tenant orgs with Twilio credentials
agent, agent_knowledge	AI agents and knowledge base links
phone_number	Twilio numbers assigned to agents
call, contact	Call records and caller contacts
analyzer, system_analyzer	Post-call analysis config
knowledge, knowledge_document	Knowledge bases for agent tools
todo, todo_type	AI-extracted to-do items
core_conf	Runtime configuration (API keys, provider settings)
report_email_address	Email report recipients

5. Generate Entities

SeaORM entity files are gitignored — they must be generated locally from the database schema:

just generate

This runs:

sea-orm-cli generate entity -o src/bases/db/schemas --with-serde both

Danger

Skipping this step causes build failures with missing module errors. Every developer and CI environment must run just generate after cloning.

6. Start the Dev Server

just dev

This runs dx serve, which compiles two targets simultaneously:

• Server — native Rust binary (Axum)
• Client — WASM binary (Dioxus)

First-run compilation takes several minutes. The app is available at http://localhost:8080 once both targets finish.

Quick Reference

just dev        # Start dev server with hot reload (dx serve)
just build      # Production build (server + WASM)
just check      # Fast type-check (native only)
just lint       # Run clippy
just fmt        # Format code
just migrate    # Run pending database migrations
just generate   # Regenerate SeaORM entities from DB schema

Note

just check and cargo check only verify the native target. They won’t catch WASM compilation errors. Use just build for a full check across both targets.

Configuration at Runtime

Once running, all API keys and provider settings are managed through the core_conf database table — not environment variables. The only env vars read at runtime are DATABASE_URL and APP_HOST.

Some defaults set by migrations:

• realtime_provider → "openai" (changeable via UI)
• google_location → "us-central1"
• google_token_uri → "https://oauth2.googleapis.com/token"