Quick start
This guide takes you from a fresh account to a working table, a view and a connected AI agent in about ten minutes.
1. Create your account
- Open the home page and click Start building.
- Enter your name, email and a password. If sign-ups are closed on your server, ask an administrator for an invitation.
- The first account created on a new server automatically becomes the instance administrator.
After signing up you land on the dashboard, which lists your workspaces and recently used applications.
2. Create a workspace
A workspace is created for you during onboarding. To create another one, open the workspace switcher at the top of the left sidebar and choose Create workspace. See Workspaces & members.
3. Create a database and a table
- In the sidebar click + New → Database.
- Name it, for example
Sales. - A first table called
Tableis created with a few sample fields. Rename it by double clicking its name in the sidebar.
You can also start from data you already have: + New table → Import file accepts CSV, Excel (XLSX), JSON and XML. See Databases & tables.
4. Add fields and rows
- Click the + at the end of the header row to add a field and pick its type (text, number, date, single select, link to table…).
- Click an empty row at the bottom of the grid, or press <kbd>Enter</kbd> on a cell, to start typing.
- Press the expand icon at the left of a row to open it as a form.
5. Make a view
Views are saved ways of looking at a table. Click the view name at the top left, then Kanban, Gallery, Calendar, Form or Grid. Filters, sorts and hidden fields are saved per view. See Views, filters & sharing.
6. Connect an AI agent
- Open Settings → MCP server and click Create endpoint.
- Choose the workspace the agent may use and copy the URL.
- Add it to Claude, ChatGPT, Codex, Cursor or another MCP client, as shown in Connect an agent.
- Ask the agent: “List the tables in my Sales database and add a Status field to Deals.”
7. Group work in a project
Create a project from Projects in the sidebar, attach your database and let the agent keep notes in the project memory.
Next steps
- Build a full example with the CRM tutorial.
- Let an agent do the work in the dashboard tutorial.