Post-Ingestion Tasks - System Jobs
After you have added an external data source to your Tenant configuration, the Aisera Gen AI platform runs a RAG index on the data and then displays it via the SOR menu items.
Navigate to the SOR menu in your Aisera Admin UI and select one of the sub-menus to see the data from your enterprise application (3rd Party/external) data source.
The ingested data fields are shown in the lower section of the Tickets, Knowledge, or Serice Catalog page. For directions on how to change the mapped fields, see Field Mapping.
Data Preparation
There are post-ingestion tasks that you may need to complete before your ingested data is ready for use. These include creating data maps for custom fields, and making sure your Knowledge Articles contain the content tags you want to use.
Running System Jobs
The Aisera Gen AI platform contains many system jobs for internal and external use, including: Neural Search RAG indexing, Knowledge Article indexing, running Access Attribute Extraction jobs for User data, or running Discovery Ontology Indexing for Ticket data.
The steps for running System Jobs are all the same, but each job may require different parameters. You may need to talk with your data source administrator to gather some of these parameters.
To Run a System Job:
Navigate to Settings > System Jobs in the Asera Admin UI.
Select the Generic Jobs tab near the top of the window.
Select the + New Job button in the upper-right corner of the window.
Scroll down the Select Job list and choose the job you want to run.
Select the OK button.
If your job requires specific parameters, more configuration windows will appear.
Select OK after entering the parameters for each window.
Select OK again on the confirmation window that asks, Are you sure you want to run this job?
Navigate to Settings > Jobs to track the progress of your job.
Description of Frequently Used System Jobs
Neural Search RAG - Summary Backfill
Runs automatically every Sunday Re-indexes documents missing summaries
All other documents are untouched
If all documents already have summaries, the job completes with no changes
You can trigger the job manually anytime from the as a System Job.
Runs automatically
Application/bot Name
Data Source
Neural Search RAG Indexer
Turns knowledge chunks into Vectors - a mathematical representation of the chunks for a more efficient search. NOTE: This option requires the Private KB Fulfillment Type to be enabled in the AiseraGPT Advanced Settings tab.
Whenever any new/updated documents are added to an AiseraGPT (Agent’s) knowledge graph
Toggle use-synonyms on
(Optional) Toggle enable-incremental-updates
(Optional) Toggle reset-index (only use in rare-cases where you want to completely wipe the existing indices and start fresh).
(optional) Configure data-sources by ID (IDs can be found in the data source configuration page). Only use if you have large volumes of documents (hundreds / thousands) and only want to update specific data source(s) to prevent a long job-run.
Click OK
Neural Search Flow Indexer
Turns the description and use-cases text-bodies attached to Workflows / Hyperflows into chunks, and turns those chunks into Vectors - a mathematical representation of the chunks for a more efficient search. NOTE: This option requires Flow with No Intent (Intentless) Fulfillment Type to be enabled in the AiseraGPT Advanced Settings tab.
After adding or updating any Workflow/Hyperflow Description/Use-cases field.
Neural Search Service Catalog Indexer
Turns the description text-bodies attached to Service Catalog Items into chunks, and turns those chunks into Vectors - a mathematical representation of the chunks for a more efficient search. NOTE: Requires Service Catalog Fulfillment Type to be enabled in the AiseraGPT Advanced Settings tab. Also requires Service Catalog Data Source configured and Service Catalog Knowlege Learning job to be executed in the tenant.
When you want to make service catalog items conversationally searchable and accessible in the chat via hyperlink. or
Ontology Generation
Used to discover new entities (words / terms / acronyms) / synonyms in client-provided data (Knowledge Docs, Ticket Data) to improve RAG search accuracy. Example: If a bot user asks about "at-home fitness," but your document only uses the term "calisthenics," a basic search might miss it.
Whenever new data source content is added to an AiseraGPT (Agent).
Navigate to LLMs -> Ontology -> + New -> Ontology Generation
Provide Any Name
(Optional) Filter for specific data sources (Data Sources shown are only those which have been added to the AsieraGPT Bot in context).
Add some use-cases to give the LLM model an example of the type of content and specific terminology to search.
Ticket Learning
Job to Load, Clean, Chunk, and Embed Tickets into the Aisera Gen AI platform.
Upon Creation / Update of Ticket Data Sources.
Run using the Start button on the Data Source that contains your Tickets.
Knowledge Learning
Singular Job to Load, Clean, Chunk, and Embed Knowledge Articles into the Aisera Knowledge Graph.
Upon Creation / Update of Knowledge Learning Data Sources.
Run using the Start button on the Data Source that contains your Knowledge Articles.
Service Catalog Learning
Job to Load, Clean, Chunk, and Embed Service Catalog Items into the Aisera Gen AI platform.
When you create or update a Service Catalog.
Run using the Start button on the Service Catalog Data Source.
Ontology Jobs
The ontology for your application/bot is added when you choose the Domain (HR, IT, Medical) in the first steps of creating your application/bot.
However, you can run a job to enhance the existing ontology or create a new ontology. Follow the steps in the table above, at the right side of the Ontology row. This is not specifically a System Job, but you can track the progress in the Settings > Jobs window.
Learning Jobs
Learning jobs run automatically when your data source connection to the tenant is complete. A separate learning job runs for Tickets, Knowledge Base Articles (Knowledge), and Service Catalog Items (most often used with ServiceNow data). The Aisera Gen AI platform will ingest data from all of the fields that are pre-mapped or manually mapped.
To run a Learning Job manually:
Navigate to Settings > Data Source.
Choose the a source that's connected to your Aisera Gen AI platform.
Select the Start (Run) button at the top of the window.
Job Sequencing
If you plan to run more than one System Job sequentially, for a new or updated database, use this order:
Run Knowledge Learning against the KB data source(s) - automatic
Run Ontology Generation to discover new Entities / Synonyms
Run Neural Search RAG Indexer to turn new/updated knowledge chunks into searchable vectors.
More Tasks
In addition, you may want to:
test your bot with AI Lens
add or modify Workflows or Hyperflows
create Events with Event Studio
add Ontologies for Tickets
analyze Unresolved conversations with AI Workbench, make revisions, and iterate
research Analytics, make revisions, and iterate
customize prompts using the Prompts Studio
Last updated
Was this helpful?