Shared Datasources - PLai Framework Documentation

A Shared Datasource is a knowledge base created in one Project that can be shared with other Projects. This enables multiple teams to maintain a single source of truth and ensure consistent information across all agents in different projects.

What Makes a Datasource Shared?

Single Knowledge Base
Multiple Agents Access
Centralized Management
Consistent Updates
Team Collaboration

A datasource is “shared” when it’s assigned to multiple agents in your project. All assigned agents can:

Search the same content
Retrieve the same results
Benefit from updates
Reference consistent information

Benefits of Shared Datasources

Single Source of Truth

Update knowledge once, all agents benefit

Consistency

Ensure all agents have the same information

Efficiency

Reduce storage by avoiding duplication

Maintainability

Easier to manage and update content

Single Agent vs Shared

Single-use datasource:
- Agent A → Datasource X
- Agent B → Datasource Y (separate copy)
Shared datasource:
- Agent A, Agent B, and Agent C → Datasource X (single copy)

Benefits of Consolidation

Metric	Separate	Shared
Storage	3x	1x
Updates	Manual sync	Single update
Consistency	Risk of divergence	Always in sync
Management	Complex	Simple

Creating Shared Datasources

Step 1: Create the Datasource

Go to Datasources

Navigate to Datasources in your project

Create New

Click Create Datasource

Choose Type

Select Manual or Google Drive

Set Details

Enter name, description, and purpose

Add Content

Upload files, add URLs, or enter text

Step 2: Assign to Multiple Agents

Go to Agents

Navigate to Agents section

Edit Agent

Open an agent’s configuration

Go to Datasources Tab

Find the Datasources section

Add Datasource

Click Add Datasource and select the shared datasource

Configure Settings

Set Top K, reranking, and other search parameters

Repeat for Other Agents

Assign to each agent that needs access

Step 3: Test Across Agents

Testing Shared Datasource:

1. Test Agent A
   - Search for information
   - Verify results are relevant
   
2. Test Agent B
   - Use same search queries
   - Confirm consistent results
   
3. Test Agent C
   - Verify retrieval works
   - Check response quality

Managing Shared Datasources

Centralized Updates

When you update a shared datasource:

Update Content
All Agents Notified
Changes Available Immediately
All Responses Use New Content

Timeline:

Edit datasource (add/remove files or content)
Changes are indexed
All agents using this datasource see updates
Immediate effect on subsequent queries

Versioning

Track changes to your shared datasource:

Version History: See what changed and when
Content Timeline: Track addition/removal of resources
Update Logs: Monitor who made changes
Rollback: Revert to previous versions if needed

Monitoring Usage

Track how agents use your shared datasource:

Agent References: See which agents access it
Query Analytics: Monitor search patterns
Performance: Track retrieval speed and accuracy
Coverage: Ensure all relevant queries find answers

Project Scope Model

Datasources are created in a Primary Project and can be shared with other Secondary Projects:

Project A (Primary — creates datasource)
- Datasource X (owned here)
Project B (Secondary — can use Datasource X)
- References Datasource X
Project C (Secondary — can use Datasource X)
- References Datasource X

Access Control

Scenario	Primary Project	Secondary Project
View	✅ Full access	✅ Read-only
Edit Content	✅ Can modify	❌ Cannot modify
Edit Schema	✅ Can modify	❌ Cannot modify
Delete	✅ Can delete	❌ Cannot delete
Share Again	✅ Can reshare	❌ Cannot reshare

Role-Based Access (Within Primary Project)

Role	Create	Edit	Delete	Share
Owner	✅	✅	✅	✅
Admin	✅	✅	✅	✅
Editor	✅	✅	Own only	Own only
Viewer	❌	❌	❌	❌

Secondary Project Access

All roles in secondary projects have:

✅ Read-only access to datasource
✅ Can assign to agents
✅ Can search and query
❌ Cannot modify content
❌ Cannot modify schema

Real-World Scenarios

Scenario 1: Support Team

Company FAQ Datasource:
- Product FAQ
- Troubleshooting Guides
- Support Procedures
Used by:
- Customer Support Agent
- Technical Support Agent
- Sales Support Agent

Benefits:

All agents answer from same FAQ
Consistent support experience
Single update point for policies
New team members can access existing knowledge

Scenario 2: Company Knowledge Base

Company Datasource:
- Employee Handbook
- Policies & Procedures
- Benefits Information
- Internal Processes
Used by:
- HR Assistant Agent
- Onboarding Agent
- Employee Support Agent
- HR Reporting Agent

Benefits:

Centralized HR information
Consistent policy explanations
Automatic updates for changes
Audit trail of information source

Scenario 3: Product Documentation

Product Docs Datasource:
- User Guide
- API Documentation
- Release Notes
- Troubleshooting
Used by:
- Customer Support Agent
- Product Assistant Agent
- Sales Support Agent
- Developer Relations Agent

Benefits:

Keep documentation current
Multiple agents reference same docs
Quick updates for new releases
Consistent technical information

Best Practices for Shared Datasources

Clear Naming

Use names that clearly indicate the content and purpose

Descriptive Metadata

Document what’s in the datasource and how to use it

Regular Audits

Review which agents use each datasource

Update Schedule

Plan and communicate content updates

Documentation

For each shared datasource, document:

Purpose: What knowledge does it contain?
Coverage: What topics are included?
Update Frequency: How often is it updated?
Owner: Who maintains it?
Related Agents: Which agents use it?
Search Tips: How to find information effectively?

Content Organization

Structure: Use consistent naming and organization
Metadata: Add tags for easy categorization
Sections: Organize by topic or department
Updates: Version control and change tracking
Cleanup: Remove outdated content regularly

Performance Optimization

Content Quality: Well-structured documents retrieve better
File Size: Optimize large files for faster indexing
Specificity: Avoid redundant or overlapping content
Reranking: Enable for improved relevance
Top K: Tune based on your use case

Challenges & Solutions

Challenge: Content Divergence

Problem: Different teams create duplicate datasources with similar content Solution:

Audit existing datasources before creating new ones
Consolidate similar content
Establish naming conventions
Document datasource purposes clearly

Challenge: Outdated Information

Problem: Shared datasource becomes stale, misleading agents Solution:

Assign ownership for updates
Set update schedule and reminders
Monitor usage patterns
Archive old content clearly
Enable version history

Challenge: Conflicting Information

Problem: Multiple versions of policies or procedures Solution:

Single source of truth approach
Clear deprecation process
Version control
Update notifications
Audit trail of changes

Migration from Individual to Shared

If you currently have separate datasources for each agent:

Audit Content

Review all existing datasources for duplication

Consolidate

Merge similar content into single datasource

Update Agents

Reassign agents to shared datasource

Test

Verify all agents work correctly

Cleanup

Delete old individual datasources

Document

Record the new shared datasource structure

Next Steps

Resource Types - Learn about files, URLs, and content
Datasource Metadata - Configure metadata and properties
Types of Datasources - Understand Manual and Google Drive options
Agent Configuration - Assign datasources to agents

​What Makes a Datasource Shared?

​Benefits of Shared Datasources

Single Source of Truth

Consistency

Efficiency

Maintainability

​Sharing Architecture

​Single Agent vs Shared

​Benefits of Consolidation

​Creating Shared Datasources

​Step 1: Create the Datasource

​Step 2: Assign to Multiple Agents

​Step 3: Test Across Agents

​Managing Shared Datasources

​Centralized Updates

​Versioning

​Monitoring Usage

​Datasource Ownership & Cross-Project Sharing

​Project Scope Model

​Access Control

​Role-Based Access (Within Primary Project)

​Secondary Project Access

​Real-World Scenarios

​Scenario 1: Support Team

​Scenario 2: Company Knowledge Base

​Scenario 3: Product Documentation

​Best Practices for Shared Datasources

Clear Naming

Descriptive Metadata

Regular Audits

Update Schedule

​Documentation

​Content Organization

​Performance Optimization

​Challenges & Solutions

​Challenge: Content Divergence

​Challenge: Outdated Information

​Challenge: Conflicting Information

​Migration from Individual to Shared

​Next Steps

What Makes a Datasource Shared?

Benefits of Shared Datasources

Sharing Architecture

Single Agent vs Shared

Benefits of Consolidation

Creating Shared Datasources

Step 1: Create the Datasource

Step 2: Assign to Multiple Agents

Step 3: Test Across Agents

Managing Shared Datasources

Centralized Updates

Versioning

Monitoring Usage

Datasource Ownership & Cross-Project Sharing

Project Scope Model

Access Control

Role-Based Access (Within Primary Project)

Secondary Project Access

Real-World Scenarios

Scenario 1: Support Team

Scenario 2: Company Knowledge Base

Scenario 3: Product Documentation

Best Practices for Shared Datasources

Documentation

Content Organization

Performance Optimization

Challenges & Solutions

Challenge: Content Divergence

Challenge: Outdated Information

Challenge: Conflicting Information

Migration from Individual to Shared

Next Steps