Trigger Ingestion Pipelines from the Admin UI

The Ingestion Jobs interface allows authorized administrators to manually trigger, review, and track ingestion pipelines for a Tenant. These pipelines synchronize enterprise content and operational data with the systems used by AIW for search, retrieval, analytics, and AI-generated responses.

Administrators can trigger the following jobs:

• Knowledge articles

• SharePoint

• Confluence

• Incident records

• Landscape intelligence

Prerequisites

Before accessing the interface, verify that:

1. You have administrator access to the applicable environment.

2. You can run jobs for the tenant.

Note

Run ingestion jobs only for tenants and environments you administer.

What Is an Ingestion Pipeline?

An ingestion pipeline reads information from a source system, processes the information, and sends it to a destination where AIW can use it.

Access the Admin Console

1. Log in to the Instance.

2. Open AIW Chat on the instance.

3. Navigate to Admin > AI > AIW Configuration.

4. The interface displays Knowledge Sync with all jobs.
   Figure: Knowledge Sync
   

5. Hover on the Job card to run the job and to view it’s history.
   Figure: Job Card Details

6. Click any source to view the Jobs for the Source. The application displays Job Status page.
   Figure: Knowledge articles job status

   Field	Description
   Run	Display the date and time when you initiated the sync run. Includes a View Details link to access more information about the run.
   Status	Display the result of the sync run, such as Succeeded or Failed.
   Progress	Display the current state of the sync run, for example Completed or Failed. For more information refer Job Statuses.
   Started	Display the date and time when the sync run started.
   Updated	Display the date and time when the sync run was last updated or completed.
   Started By	Identifies what triggered the sync run, such as Manual or an automated process.
   Summary	Provides a summary of the sync run, including the number of records found and processed, the number of chunks created, and the number of rows ingested.

7. Click view details on the run time to view the full job details.
   Figure: Run Details

   Field	Description
   Source	Display the name of the Data source associated with the ingestion run.
   Progress	Display the current state of the ingestion run, such as Completed, In Progress, or Failed. For more information, refer Job Statuses
   Started By	Identifies how the ingestion run started, for example, Manual or Scheduled process.
   Started	Display the date and time when the ingestion run began.
   Updated	Display the date and time when the ingestion run was last updated or when processing completed.
   Completed	Display the date and time when the ingestion run finished.
   Summary	Provides a summary of the ingestion run, including the number of items found, successfully processed items, generated chunks, and files ingested.
   Status	Display the total result of the ingestion run, such as Succeeded or Failed.

Note

Before running a job, always verify the environment and tenant. Running a full-refresh job for the wrong tenant can replace that tenant’s existing indexed content.

Run an Ingestion Job

1. Log in and navigate to AIW configuration to view the Knowledge Sync Jobs page.

2. Click Run now by hovering on the job card. When the job starts running, the system displays Running as an action until the job completes.

3. Track the status until it changes to Succeeded or Failed.
   Figure: Running Job

Note

Most jobs run inside the agent process but, for Landscape Intelligence jobs run remotely.

After triggering a job, use the status displayed in the UI to determine its progress. A successful run ends with Completed .

Available Jobs

Knowledge Articles

The Knowledge Articles job synchronizes Apex Enterprise knowledge articles.

After the job completes, AIW can retrieve these articles when answering user questions.

To run the Knowledge Articles job, perform the following steps:

1. Select Knowledge Articles.

2. Select Run now.

3. Track the job until it reaches Succeeded or Failed state.

The tenant’s knowledge articles become searchable through AIW’s RAG Workflow.

SharePoint

The SharePoint job synchronizes files from a SharePoint or OneDrive drive.

To run the SharePoint jobs, perform the following steps:

1. Select SharePoint.

2. Select Run now.

3. If supported by the UI, provide a root folder or individual file identifier.

4. Track the job until it reaches Succeeded or Failed state.

Supported SharePoint and OneDrive documents become searchable in AIW to fetch the details.

Confluence

The Confluence job synchronizes Confluence wiki content. This allows AIW to search and retrieve Confluence information as enterprise knowledge.

1. Select Confluence.

2. Verify that you selected the correct tenant.

3. Select Run now.

4. Track the job until it reaches Succeeded or Failed.

5. Review the completion summary, when available.

While the UI displays a running state, the pipeline can be performing one or more of the following operations:

1. Discovering content: The pipeline crawls configured Confluence spaces or retrieves a specific page. It counts the discovered pages, attachments, and images.

2. Removing outdated content: The pipeline deletes existing content before it indexes the new content. This is a full-refresh behavior for Confluence content.

3. Preparing content: The pipeline retrieves page HTML and supported attachments, identifies semantic structures, and divides the content into searchable chunks.

4. Indexing content: The pipeline inserts or updates content.

5. Completing the job: The pipeline records processing statistics and the final job status.

What gets ingested

The Confluence pipeline can ingest:

• Classic Confluence pages

• Blog posts

• Headings, Paragraphs, Tables, Ordered and unordered lists, Code sections, Panels, Expandable sections and Tasks

• Supported file attachments and PDFs

• Supported images through OCR

The ingestion process skips unsupported or non-text media, such as ZIP archives and video files.

AIW retrieves the indexed copy rather than calling Confluence for every chat message.

Incident Records

The Incident Records job ingests closed Apex incidents. The indexed records can support similarity searches, analytics, AIW features, and landscape intelligence.

The job runs inside the agent process. Command-line and scheduled Multi-Tenant execution paths can also be available outside the Admin UI.

To run the Incident Records job, perform the following steps:

1. Select Incident Records.

2. Select Run now.

3. If the UI exposes a mode or time-window parameter, select the applicable incremental or backfill option.

4. Track the job until it reaches Succeeded or Failed state.

Historical incident vectors become available for similarity analysis and downstream landscape intelligence processing.

Landscape Intelligence

Landscape intelligence analyzes incident information to produce clusters, keywords, labels, and reports.

Unlike the other jobs, this is not primarily a document ingestion pipeline. The agent sends a trigger request to the Landscape Intelligence service, which starts a separate worker job.

Landscape trigger→ remote AI-Predictive service and worker

The worker generates analytical outputs in Landscape Intelligence.

Note

The main processing runs remotely in the Landscape Intelligence service.

To run Landscape Intelligence job, perform the following steps:

1. Select Landscape Intelligence.

2. Select Run now.

3. Track the extended analysis stages.

4. Wait for the remote worker to report Succeeded or Failed.

What happens during the job

1. The agent records INITIATED in Couchbase.

2. The agent sends the tenant, job, run, mode, and trigger information to the Landscape Intelligence API.

3. The API starts a worker job.

4. The worker processes the available incident information.

5. The worker clusters incidents.

6. The system generates keywords and labels.

7. The worker saves the results.

8. The system generates and publishes Landscape reports.

9. The worker updates the shared job status.

The agent does not mark the job as completed. The remote Landscape Intelligence worker handles the later stages and final status.

Landscape intelligence generally depends on incident data already being available.

The workflow produces clusters, keywords, labels, and reports. Landscape Intelligence stores these outputs in its data stores or Couchbase rather than following the standard document-ingestion model used by Confluence, SharePoint, and Knowledge Articles.

Job Comparison

Job Statuses

The UI converts internal pipeline stages into administrator-friendly status labels.