Documentation

    Run Byaan locally or deploy the Team Version

    The Mac app is free for individuals and runs entirely on your machine. The team version is one-line Docker — multi-user auth, RBAC, Slack, Google OAuth, and automatic HTTPS, all in your VPC.

    Quick start

    01

    Download Byaan for macOS.

    02

    Connect a database or upload CSV, Excel, Parquet, or JSON files.

    03

    Let Byaan inspect schemas and learn your business definitions.

    04

    Ask questions in natural language and turn answers into dashboards.

    Core capabilities

    Connect to your data

    PostgreSQL, MongoDB, MySQL, MariaDB, SQLite, MSSQL, DuckDB, plus CSV, Excel, Parquet, and JSON files.

    Ask in plain English

    Byaan converts business questions into read-only analysis and preserves your metric definitions and table context.

    Build dashboards

    Generate interactive dashboards with filters, charts, and reusable analysis patterns from notebooks.

    Use MCP

    Expose Byaan as an MCP server for Claude Code, Cursor, Codex, and other MCP-compatible clients.

    Read-only by design

    DDL and DML are blocked. Production databases stay safe — Byaan can only inspect, never write.

    Export and share

    Export standalone HTML dashboards, share notebook links, or generate PDFs for stakeholders.

    Team Version

    Full README →

    One container ships PostgreSQL, the FastAPI backend, the React frontend, and Caddy. Live on port 8080 in five minutes. AGPL licensed.

    One-line install
    curl -fsSL https://downloads.byaan.ai/docker/install.sh | bash

    Drops start.sh and .env into a byaan directory. Set admin email, password, and org name in .env, then run ./start.sh.

    Requirements

    • Linux server (Ubuntu 20.04+, Debian 11+, or any Docker-compatible OS)
    • 2 GB RAM minimum (4 GB recommended), 10 GB disk
    • Ports 80 and 443 (or 8080 for local)
    • Docker installed

    Required environment variables

    VariableDescription
    APP_SECRETJWT signing + encryption key (auto-generated by installer)
    MASTER_USER_EMAILAdmin user email
    MASTER_USER_PASSWORDAdmin password (min 8 characters)
    ORG_NAMEYour organization name

    Optional

    VariableDescription
    ENCRYPTION_KEYSeparate 32-byte hex key for credential encryption (recommended)
    DOMAINCustom domain (e.g. app.yourcompany.com) — enables Let's Encrypt HTTPS
    ACME_EMAILEmail for Let's Encrypt cert notifications
    GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRETGoogle OAuth (optional SSO)
    SMTP_HOST / SMTP_USERNAME / SMTP_PASSWORDSMTP for team invites (optional)

    Commands

    • ./start.shStart Byaan
    • ./start.sh stopStop Byaan
    • ./start.sh updateUpdate to latest version
    • ./start.sh logsView all logs
    • ./start.sh logs backendView backend logs only
    • ./start.sh logs caddyView web server logs
    • ./start.sh logs postgresView database logs
    • ./start.sh statusCheck if running
    • ./start.sh removeRemove container, keep data
    • ./start.sh remove --dataRemove container and all data

    Backups

    Data lives in the byaan_data Docker volume. Snapshot via pg_dump or volume tar.

    Team invites

    Configure SMTP to email invitations automatically. Or skip SMTP — invitations are still created and the admin copies the generated link from the Members page to share manually.

    Updates & logs

    Keep your Team Version current with a single command. Data and configuration are preserved across updates. Use the log commands to monitor or debug any service.

    Zero-downtime updates

    ./start.sh update pulls the latest image, recreates the container, and mounts the same byaan_data volume — no data loss.

    Targeted log tailing

    Pass a service name to scope output: backend, caddy, or postgres. Omit it to follow all services at once.

    Security model

    The Mac app keeps database credentials, queries, and result data on your computer. You choose the model provider and keys, and Byaan connects from your environment.

    The team version runs in your VPC. Byaan-hosted infrastructure is never in the data path — credentials, queries, and results stay where you put them.

    Both versions are read-only by design: DDL and DML are blocked, so production databases can be inspected without risk of accidental writes.

    Have a question?

    Open an issue or jump into discussions on GitHub.