Skip to main content

Knowledge Sources

Create a Knowledge Source, choose data and fields, and verify retrieval from an AI Agent.

1. Overview

Knowledge Sources in Yeeflow provide a centralized knowledge management layer designed to organize and deliver structured business information to AI-driven applications.
They allow administrators to consolidate documents, operational rules, reference materials, and structured datasets into a unified knowledge foundation that can be consumed by AI Agents.

A Knowledge Source defines inputs for the agent. Test answers against known records before publishing, and check the source again when its filters, fields, or data change.

Typical use cases include:

  • Providing onboarding guides or reference materials for new employees

  • Offering SOPs (Standard Operating Procedures) for approval workflows or data list operations

  • Storing business rules, policy documents, and FAQ collections

  • Maintaining process descriptions, field definitions, and data model documentation

  • Sharing reusable knowledge across departments to ensure consistent operations

A Knowledge Source can combine sources such as a list of SLA targets and a document library of support policies. Select the records and fields the agent needs from each source.


2. Key Benefits

Feature

Description

Group related sources

Combine data lists and document libraries that answer related questions.

Choose available fields

Include the fields the agent needs. Available fields depend on the selected source.

Reuse a configured source

Add the same resource to each AI Agent that needs it.

Limit records

Use filters to include the records relevant to the task.

Find an existing resource

Search by resource name when attaching a source to an agent.

Follow data changes

Enable Automatic sync when needed, then verify updates with a known record.


3. Accessing Knowledge Sources

Knowledge Sources are configured at the application level.

Navigate to:

Application Settings → AI agent → Knowledge sources

If no resources exist, the page displays a prompt to create your first Knowledge Source.


4. Creating a Knowledge Source

Click Create resource to open the configuration window.

4.1 Basic Information

Field

Description

Name

Required. The name of the Knowledge Source

Description

Optional. A high-level description of the data scope and how the included data sources relate to each other.

Custom icon

Optional. Upload a custom image


5. Managing Knowledge Sources

After creation, the resource appears in the list with management options:

  • Edit resource

  • Disable resource

  • Delete resource


6. Adding Data Sources to a Knowledge Source

A Knowledge Source can include multiple underlying datasets, such as:

  • Data lists

  • Document libraries

Open a Knowledge Source → click Add data source.


6.1 Configuring a Data Source

When adding a data source, complete the following steps:

Step 1 — Select Data

  • Choose the Application

  • Choose the Data source (data list or document library)

  • (Optional) Apply filter conditions

  • Enable Automatic sync to keep the data updated

Step 2 — Select Data Fields

Select the fields needed for the agent’s task. For a support-policy source, select the policy name, status, and target response time if those fields are needed for the answer. Leave unrelated fields unselected.

Select data showing the Application, Data source, Set filter conditions and Automatic sync controls.

Click OK to confirm.


6.2 Managing Data Sources

Each added data source can be:

  • Edited (modify fields, filters, sync settings)

  • Deleted (removed from the Knowledge Source)


7. Disabling or Deleting a Knowledge Source

You may:

Disable resource

  • Makes the Knowledge Source temporarily unavailable

  • Can be re-enabled anytime

Delete resource

  • Permanently removes it

  • Cannot be undone


Knowledge Source Description and Data Scope

The Description of a Knowledge Source explains what data is included and how the data sources work together.
It provides context for administrators and AI Agents, helping both understand the scope and intent of the knowledge being configured.

Rather than describing a single dataset, the Description should summarize the combined data sources and their relationships within the Knowledge Source.

What to include in the Description

When writing a Description, focus on the following aspects:

  • The types of data sources included
    (for example, core records, reference data, configuration data, or activity records)

  • How these data sources are related
    (such as reference relationships, classifications, or mappings between datasets)

  • What the combined data enables the AI Agent to understand or answer
    (the overall knowledge capability, not individual fields)

The Description should reflect the Knowledge Source as a whole, not just one data list.

Describing multiple data sources

If a Knowledge Source contains multiple data sources, the Description should explain how they form a unified knowledge context.

Instead of listing each data source individually, describe their collective role.
For example, explain how one data source provides core records, while others provide classifications, definitions, or supporting context.

This helps clarify why multiple datasets are grouped together and how they complement each other.

Focusing on relationships, not implementation

The Description does not need to explain technical details or field-level configurations.
Instead, it should describe relationships at a conceptual level, such as:

  • How records reference or depend on each other

  • How classification or configuration data enriches core records

  • How related datasets enable more accurate interpretation

This keeps the Description concise and aligned with the purpose of the Knowledge Source.


8. Using Knowledge Sources in AI Agents

Once Knowledge Sources are configured, they can be connected to AI Agents.


Step 1 — Open the Knowledge Tab

Navigate to:

AI Agent → Knowledge → Add Knowledge

In Select resource, browse the application or search for the resource by name. Select an application from the dropdown if you need to change the scope shown.


Step 2 — Select Knowledge Resources

Examples of commonly used resources:

  • SLA targets

  • Ticket activity records

  • Service categories

  • Configuration items (CMDB)

  • Support team structure

  • Request types

Select resource showing the Search knowledge field and the Add button for an existing Knowledge Source.

In Select resource, find the existing Knowledge Source and click Add next to its name. Repeat for any other source the agent needs, then confirm each one appears on the Knowledge tab.


Step 3 — Enable or Disable Resources

For each resource:

  • Enable → The agent can query the data

  • Disable → Temporarily hides the data from the agent

Check the Enabled switch for each attached resource, then use Preview & Test Run to check retrieval. Confirm the answers use the intended records and fields.


Step 4 — Customize Resource Name & Description (Optional)

Within an AI Agent, you may:

  • Rename the Knowledge Source

  • Edit its description

These changes only affect this specific agent.
The global Knowledge Source configuration remains unchanged.

Use this when:

  • The same dataset serves different business contexts

  • You want more user-friendly names for agent use

  • You need to provide agent-specific instructional descriptions


9. Best Practices

  • Use clear, descriptive names for each Knowledge Source

  • Group related datasets into a single Knowledge Source

  • Select only the fields the agent needs. In Preview & Test Run, check a record with a known value, a record excluded by the filter, and a field you left out. Repeat from the intended entry point with representative test accounts.

  • Keep descriptions specific and actionable

  • Enable automatic sync where the agent needs to follow changes. Change a harmless test value and ask the same question after sync to verify that the agent returns the new value.

  • Disable unused Knowledge Sources to reduce noise


10. Common Use Cases

Customer Support

  • Query SLA targets

  • Retrieve ticket histories

  • Access troubleshooting articles

IT Operations

  • Access CMDB and configuration records

  • View asset details

  • Review service dependencies

Inventory / Logistics

  • Query product inventory

  • Check warehouse stock

  • Track item locations

HR / Internal Services

  • Request categories

  • Policy libraries

  • SOP collections


11. FAQ

Q: Can one Knowledge Source contain multiple data lists?

Yes. Each Knowledge Source may include multiple lists or libraries.

Q: Will disabling a Knowledge Source remove it from all AI Agents?

It becomes unavailable but not removed.
Re-enable anytime.

Q: Can I change the icon after creating the resource?

Yes, through Edit resource.

Q: How often does data sync?

Automatic sync follows changes when enabled for the data source. If results are stale, check the source configuration and its last sync time when shown, then test retrieval again.

Did this answer your question?