Use the Aisera package to upload files to an Aisera data source for indexing and to ask questions against ingested content from within your automations.

The Aisera package lets you integrate Aisera knowledge services into your automations. You can upload documents to Aisera data sources so they are indexed and available for retrieval, and then query that ingested content to retrieve answers and citations that your automation can use in downstream processing steps.

The package contains the following actions:

  • Ingest Content: Uploads a file to an Aisera data source or app so the content is indexed and available for retrieval. See Ingest Content action.
  • Ask Question of Content: Sends a natural-language query to Aisera against indexed content and returns the top answer with citation details. See Ask Question of Content action.

Prerequisites

Before you use the Aisera package actions, confirm that you have the following:

  • A Control Room OAuth connection configured and available. The Ingest Content action uses this connection to authenticate requests to the Aisera API.
  • The Aisera tenant base URL for content ingestion requests. For example, https://<your-tenant>.api.aisera.cloud.
  • The Aisera chat base URL for question requests. For example, https://<your-tenant>-<port>.chatbot.aisera.cloud.
    Important: The chat base URL used for the Ask Question of Content action is different from the tenant base URL used for the Ingest Content action. Confirm both URLs with your Aisera administrator.
  • A data source or Aisera GPT App authorization token for the Ingest Content action. To get this token, open the App details page in your Aisera tenant, navigate to Data Source, double-click the data source you want to use, and copy the Token value from the data source details page.
  • An API bearer token from the Channel details page in your Aisera tenant for the Ask Question of Content action.
  • An agent channel ID configured in Aisera for the Ask Question of Content action. The channel ID appears on the Channel Details page in your Aisera tenant.
  • A file and a unique file identifier to use when ingesting content. The unique file identifier is a stable reference: reusing the same identifier in a subsequent ingest request updates or replaces the previously indexed content for that identifier.

Supported file types

The Ingest Content action supports the following file types. Audio and video files are not supported.

  • PDF
  • TXT
  • DOC, DOCX
  • XLS, XLSX
  • CSV
  • JSON
  • HTML
  • PPTX
Important: The file picker does not filter by supported file type. If you select an unsupported file type, the Aisera API returns an error. Use the Error output variable in your automation to capture and handle the error message.

Limitations

  • This package is available in Task Bot automations only.
  • Only the Control Room OAuth connection authentication option is supported for content ingestion.
  • The Ingest Content action fails if the Unique file ID field is empty.
  • The Ask Question of Content action accepts the Metadata JSON string field value only as a valid JSON object string.
  • If the selected file cannot be read as a stream, the Ingest Content action falls back to using the local file path. The action fails if the local file path is empty or the file does not exist at that path.