Collections Management
Collections group related tables so your workflows and agents only reference what's relevant.
They help you scope AI retrieval, keep tables organized, and see at a glance which sources feed which use case. Collections are optional — they matter most once you have more than a handful of tables.
When to Use (and Not Use) Collections
Use a Collection when:
- You want to limit retrieval to a specific topic or knowledge domain
- A workflow or agent should only reference a defined set of tables
- You're managing multiple tables that belong to one process, team, or audience
Examples:
- Grouping SOP and policy tables for HR or compliance workflows
- Creating a collection of onboarding guides, templates, and customer FAQ tables
- Organizing financial tables for budgeting, forecasting, or spend approvals
You may not need a Collection when:
- You're working with just one or two general-purpose tables
- The table is referenced across many unrelated workflows
- You're testing or uploading content for a short-term use case
In those cases, using a table directly — without grouping it into a collection — is often faster.
What Collections Can Contain
Collections group tables regardless of how they were created:
Type
Examples
Native tables
Tables you built directly in MergeBase with a custom schema
Uploaded tables
Parsed from CSV, Excel, PDF, Word, or audio files
Connected tables
Synced from Salesforce, HubSpot, Google Sheets/Drive, Notion, Postgres, and other connected sources
Example Collection Use Cases
Collection Name
Purpose
Sample Tables
Customer Resources
Central hub for support and onboarding content
Help docs table, email templates, CRM export
Sales Analytics
Revenue performance and forecasting data
Salesforce sync, pipeline table
Product Library
Product specs and documentation
Feature docs, release notes table
What You'll See in the Interface
Each table shows its collection membership, along with:
- Table name and row/record count
- Source type (native, uploaded, or connected)
- Last sync time and status, for connected tables
- Workflow usage — where the table is referenced
Best Practices
Organize Clearly
- Group by function, team, or workflow
- Use descriptive collection and table names
- Keep a narrow scope per collection to avoid retrieval noise
Maintain Proactively
- Review and clean up collections periodically
- Remove or archive tables that are no longer used
- Monitor which workflows depend on which tables before removing or restructuring one
Was this helpful?