Building Your First Lark Base: A Practical Ops Guide

Most Lark users start with messaging and calendar. They pick up Docs within the first week. But Lark Base — the platform's relational database module — tends to stay unopened for months, because it looks more technical than it actually is. That is a significant missed opportunity. A well-built Lark Base can replace your project tracking spreadsheet, your client CRM, your onboarding checklist, and your approval workflow — all in one place, connected to everything else in Lark.

This guide will take you from a blank Base to a functional, linked, automated operational database. We will use a project tracker as the example throughout, because it translates directly to almost any team's real needs.

What is Lark Base?

Lark Base is a no-code relational database builder, similar in concept to Airtable but with tighter integration to the rest of the Lark platform. A Base contains one or more tables. Each table is made up of fields (columns) and records (rows). Fields can be text, numbers, dates, dropdowns, checkboxes, attachments, formulas, or links to records in other tables. You can layer different views on top of the same data — a grid, a Kanban board, a calendar, a gallery — without duplicating anything.

Where Lark Base becomes particularly powerful is in its automation engine and its native connections to Lark Messenger, Lark Approval, and Lark Calendar. A status change in a Base record can send a Lark message, create a calendar event, or trigger an approval flow — without any third-party integration tool.

Step 1: Create your base and first table

1. In Lark, click the grid icon in the left sidebar to open the navigation panel, then select "Base" and click "New Base."
2. Give your Base a name — for this guide, call it "Project Tracker."
3. Lark will create a default table called "Table 1." Rename it by double-clicking the tab at the top of the table view. Call it "Projects."
4. You now have a blank table with a single default "Name" field. This is your starting point.

Before you start adding fields, spend two minutes thinking about what records this table will hold. In this case, each row represents a project. Keep the scope of a single table narrow — one table, one type of thing. You will link to other tables for related data.

Step 2: Set up your field types

Click the "+" icon to the right of the last column header to add new fields. For a project tracker, set up the following fields:

1. Project Name — the default "Name" field; this is a text field and serves as the primary identifier for each record.
2. Status — choose field type "Single Select." Add options: Not Started, In Progress, In Review, Completed, On Hold. You can colour-code these for the Kanban view later.
3. Owner — choose field type "Person." This links to Lark user accounts on your workspace, so you can assign projects to team members directly.
4. Start Date and Due Date — choose field type "Date" for each. Enable "Include time" only if you need it; for most project trackers, date-only is cleaner.
5. Priority — another "Single Select" field with options: High, Medium, Low.
6. Notes — a "Long Text" field for any context that does not belong in a dedicated field.

Save each field by clicking the checkmark or pressing Enter. You can always add, remove, or reorder fields later without losing data.

Step 3: Choose your view

The default Grid view works like a spreadsheet and is the most versatile for data entry. But Lark Base lets you create multiple views of the same table, each with its own filters, sort order, and visible fields. Click "+ Add View" in the left panel to see the options.

• Grid — the default spreadsheet layout. Best for bulk editing, importing data, and seeing everything at once.
• Kanban — groups records by a Single Select field. Set it to group by Status and you have an instant project board. Drag cards between columns to update status.
• Gallery — displays records as cards, useful when you have attachments or images to show.
• Calendar — plots records against a date field. Set it to use Due Date and you can see what is due when across all projects at a glance.
• Gantt — creates a timeline view using Start Date and Due Date. Essential for teams managing dependencies across projects.

Create a Kanban view grouped by Status. This will be the view your team uses for daily standups. Then create a Calendar view using Due Date. Switch between them using the view tabs on the left — the underlying data is identical, only the presentation changes.

Step 4: Link tables together

The real power of a relational database comes from linking tables. Create a second table in your Base called "Tasks." This table will store individual tasks, each linked to a parent project in the Projects table.

1. In the Tasks table, add a field of type "Link to Another Record." Select the Projects table as the source.
2. Name this field "Project." Now each Task record can be linked to one or more records in the Projects table.
3. When you link a task to a project, Lark automatically creates a reverse link in the Projects table, so you can see all related tasks from within a project record.
4. Back in Tasks, add a "Lookup" field. Choose the "Project" link field as the source, then select "Status" from the Projects table. This will pull the parent project's status into each task row automatically — no manual updating required.

Add the usual fields to Tasks: Task Name, Assignee (Person field), Due Date, Status, and a "Notes" long text field. You now have a two-table system where tasks are always linked to their parent project and can surface project-level information without duplication.

"A linked table is not just a foreign key — it is a live connection. When the project status changes, every lookup field that references it updates automatically across every task record."

Step 5: Add a simple automation

Automations in Lark Base are set up via the "Automation" panel, accessible from the toolbar at the top of the Base. Each automation has a trigger and one or more actions.

1. Click "Automation" in the top toolbar, then "New Automation."
2. Set the trigger to "Record field changes." Select the Projects table and the Status field as the trigger source.
3. Add a condition: only run this automation when Status changes to "Completed."
4. Add an action: "Send a Lark message." Choose a Lark group chat or an individual user as the recipient. Write a message template using dynamic fields — for example: "Project [Project Name] has been marked as Completed by [Owner]."
5. Save and enable the automation.

Now, whenever anyone changes a project's status to Completed, a message is automatically sent to the designated chat. No manual updates, no missed notifications. You can add additional actions to the same trigger — create a calendar event, send an email via Lark Mail, or trigger a Lark Approval flow for a sign-off.

Real-world example: a simple project tracker

Combining everything above, here is what a minimal but functional project tracker looks like in practice:

• Projects table — one record per project, with Status, Owner, Priority, and date fields. Kanban view for the team board, Calendar view for deadline visibility, Grid view for bulk updates.
• Tasks table — one record per task, linked to a parent project. Grid view filtered by Assignee (each team member has a personal view showing only their tasks). Sorted by Due Date ascending.
• Automation 1 — when a Project Status changes to "In Review," send a message to the review channel with the project name and owner.
• Automation 2 — when a Task Due Date is tomorrow and Status is not "Completed," send the Assignee a direct reminder message.

This setup takes about 45 minutes to build from scratch if you have not used Lark Base before. Once it exists, it replaces a project spreadsheet, a task list in another app, and the manual "just a reminder" messages that clog up most team chats. As your needs grow, you add tables — a Clients table, a Deliverables table, a Budget table — and link them in. The system scales with you rather than against you.

The hardest part of building in Lark Base is not the tool itself — it is resisting the urge to add every field you can think of in the first session. Start with the minimum viable structure, use it for two weeks, then add what you actually find yourself needing. Iterating on a working system is always faster than building a perfect one upfront.

← Back to blog