Data Sources & Sync
Your knowledge base is only as effective as the data behind it. MergeBase supports two ways to get data into a table — uploading a file or connecting a source — and keeps connected tables fresh with scheduled or on-demand sync. This guide explains how each works and how to manage sync health.
Source Types Overview
Source Type
Description
Sync Behavior
Uploaded files
CSV, Excel, PDF, Word, or audio
One-time parse; re-upload to refresh
Connected sources
Databases and SaaS apps via 30+ connectors (Salesforce, HubSpot, Google Sheets/Drive, Notion, Airtable, SharePoint, Postgres, Snowflake, Supabase, Outlook, Teams, Gmail, and more)
Scheduled or on-demand sync, Incremental or Full
Adding a Data Source
- Click "Create Table" in the Knowledge Base header
- Choose an option from the dropdown:
- New table — define your own schema
- Upload file — parse a file into a table
- Connect source — link a database or app
- Configure the connection — provide credentials, or authenticate via OAuth for the connector
- Choose what to sync — select the specific tables, sheets, or objects to bring in
File Upload Details
Supported Formats PDF, CSV, Excel, Word, and audio (validated during upload)
Upload Interface Features
- Drag-and-drop zone
- Manual file browser
- Multi-file upload
- Real-time progress indicators
- File type validation
Sync Status Reference
Status
Meaning
Succeeded
The sync completed and data is current
Running
A sync is in progress right now
Pending
A sync is queued and hasn't started yet
Failed
The sync failed — check connector credentials or error details
Re-syncing a Table
To manually refresh a connected table:
- Click Sync now next to the table
- Choose a sync mode:
- Incremental — pulls in only new or changed records since the last sync (recommended for routine refreshes)
- Full — rebuilds the entire table from the source (use after a schema change or when Incremental sync isn't catching updates)
Sync progress is shown in real time; a Failed status surfaces the underlying error so you can fix credentials or connector configuration and retry.
Source Management
To ensure stable and high-quality retrieval, follow these best practices:
- Regularly check sync status, especially for tables tied to active workflows
- Monitor which workflows reference each table to understand the impact of a sync failure
- Group related tables into Collections for better organization
- Use clear, consistent naming conventions for tables and connections
- Remove unused or duplicate tables to reduce clutter
- Configure field roles (Vectorize / Filter / Display / Ignore) after every schema change — see the AI Optimization guide
Was this helpful?