Skip to main content

Getting started: Overview

  • The Getting Started section takes you from signing up for Quill to a chat widget on your site, where your users ask and your AI agent answers from Quill's live copy of your SQL data.
    This page presents the Getting Started steps and the things to prepare before you start.

  • The configuration steps run in wizards that save you the manual work: Quill drafts the mapping of your data, the agent, and the agent's queries, and you approve or adjust each draft.

  • The chat widget is one of several channel types Quill offers, all carrying conversations between your users and an agent. You can also add a Telegram bot, a Slack app, or a Discord bot.

  • Terms like mirroring, agents, and channels are explained on the Overview page.

  • No separate server is needed to try Quill locally: you can take all the steps on your own machine, with your Quill domain pointing at 127.0.0.1.

  • In this article:

Prerequisites​

Before taking the first step, make sure you have:

  • An email address, to receive Quill's access details.

  • A machine that will run Quill.

    • The machine needs Docker installed: Docker Desktop on Windows and macOS, or Docker Engine on Linux.
    • Port 443 must be available: the dashboard, the API, and the chat channels of your deployment are all served on this one port.
    • Quill's machine needs internet access: Docker pulls Quill's image from Docker Hub, and Quill fetches the certificate and configuration for your domain from api.ravendb.net.
    • The machine's public IP address: your Quill domain will point at this address.
      To try Quill locally, use 127.0.0.1.

  • Your source database: PostgreSQL, SQL Server, or MySQL.

    • Quill must be able to reach the database from inside its Docker container.
    • The database must be recording its changes. Quill turns change recording on where it can.
      For the conditions each database type must meet, see Connecting your database.
    • The credentials of a database user: Quill connects to the source database as this user.

  • The access details for an LLM provider.
    Quill connects to OpenAI, Azure OpenAI, or Ollama using access details that you provide, e.g., your OpenAI API key.
    Your agent will use the LLM in every conversation it handles.

The Getting Started steps​

The first step runs on the sign-up page, the second in a terminal on your machine, and the rest in Quill's management dashboard.

  1. Signing up
    On the sign-up page, name your new Quill deployment, and receive the access keys and the Docker command that starts it.
    Your deployment has a name, a license key, and a dashboard API key.

  2. Starting Quill
    Run the Docker command in a terminal on your machine.
    Quill is running on your machine, and its management dashboard is open.

  3. Connecting your database
    In the management dashboard, start the Add new app wizard: point Quill at your source database and select the tables to read from.
    Quill can reach your source database, and the tables your app will work with are selected.

  4. Mapping your tables
    Continue through the same Add new app wizard: settle the mapping, by which Quill builds its internal database documents from your tables, and create the app.
    Your app is created, and Quill keeps its internal database current with your source data.

  5. Adding an AI agent
    In the Add agent wizard, connect Quill to your LLM provider, pick a drafted agent or configure one of your own, and chat with the draft agent before saving it.
    Your app has an agent that answers from its internal database.

  6. Adding a chat widget
    Create a Web widget channel for the agent, generate an embed link for the channel, and place the widget on a page of your site.
    Your users can reach the agent through a chat widget on your website.