Skip to main content
Version: Current

Knowledge Grounding

Overview​

Knowledge Grounding enables a Conversational Agent to answer questions using enterprise knowledge that has been ingested into the platform.

Instead of relying solely on the underlying AI model, the agent can retrieve relevant information from configured knowledge sources and use that information when generating responses.

Knowledge grounding helps improve:

  • Response accuracy
  • Business relevance
  • Consistency
  • Knowledge discovery
  • User trust

Knowledge Grounding Workflow​

Knowledge grounding in IB-X begins with knowledge ingestion.

Before a Conversational Agent can use enterprise content, the content must first be configured and ingested through the Knowledge Ingestion Service.

The overall process is:

Configure Knowledge Sources
           │
           ▼
Knowledge Ingestion Service
           │
           ▼
Knowledge Processing
           │
           ▼
Conversational Agent
           │
           ▼
Grounded Responses

Configuring Knowledge Grounding​

Knowledge grounding is configured at the Agent level.

To associate enterprise knowledge with a Conversational Agent:

  1. Open the Agent in the Agent Designer.
  2. Click Ingestions from the bottom of the designer canvas.
  3. Configure and ingest the required knowledge sources.
  4. Wait for the ingestion process to complete.

Once the content has been successfully ingested, it becomes available to the Conversational Agent during conversations.

Knowledge Ingestion Service​

Knowledge grounding relies on the Knowledge Ingestion Service.

The Knowledge Ingestion Service is responsible for configuring, ingesting, processing, and maintaining the knowledge sources used by Conversational Agents.

For detailed information about:

  • Supported source types
  • Source configuration
  • Website ingestion
  • File upload ingestion
  • Administration
  • Operations
  • Troubleshooting

refer to the Knowledge Ingestion Service documentation.

Using Knowledge During Conversations​

Once knowledge sources have been ingested, the Conversational Agent can use the available knowledge when responding to user questions.

Typical scenarios include:

  • Product support
  • Employee assistance
  • Policy lookup
  • Documentation search
  • Knowledge discovery
  • Internal helpdesk solutions

The agent automatically retrieves relevant information from the configured knowledge sources and incorporates that information into its responses.

Maintaining Knowledge​

Enterprise knowledge changes over time.

To keep responses accurate:

  • Regularly review configured knowledge sources.
  • Re-ingest content when source information changes.
  • Remove obsolete content when no longer required.
  • Validate knowledge coverage before publishing agents.

Well-maintained knowledge sources lead to higher-quality responses and improved user experience.

Best Practices​

  • Ingest authoritative content only.
  • Keep documentation current and accurate.
  • Organize knowledge sources logically.
  • Use descriptive source names.
  • Regularly monitor ingestion status and processing results.
  • Test knowledge retrieval using representative business questions.