Skip to main content
Version: 10.3

Implement Analytics for AI Agents

ODF 2 provides a standardized extensible Data Model for all AI Agent implementations. The model includes a set of tables with dynamic, convention-based names that enable ETL (extract, transform, load) processes to manage data consistently. All predefined columns in these tables are automatically processed by WorkFusion Analytics.

Advantages of standardized Data Model​

The standardized Data Model offers the following benefits:

  • Optimizing data flows by moving predefined and well-structured data views to the WorkFusion Analytics data warehouse.

    The source data is stored in the ds schema and serves as the operational layer for process execution. Analytics ETL processes move transformed and structured data to a dedicated data warehouse. Thus, data access patterns are isolated, with reporting and Analytics read access moved to a separate storage. This minimizes impact and dependencies for actual process executions, enabling simultaneous data read and insert operations.

  • Providing unified and reusable Analytics to cover multiple AI Agents.

  • Utilizing AI Agent-specific data for building custom Analytics dashboards based on out-of-the-box capabilities.

    The Data Store tables defined within the standardized model already include core fields required for the most common analytical needs. However, you can extend the set of core fields using the _bi custom attribute in the following format: [attribute name]_bi.

    Subsequently, the ETL engine automatically processes and moves the custom _bi columns to Analytics. This simplifies further customization and allows one to include AI Agent-specific data and attributes.

Developers are not limited to the list of tables defined under the standardized, extensible Data Model. However, any additional table created cannot be automatically processed by ETL for Analytics purposes and further usage in BI tools.

ETL extension​

The ETL extension to cover the standardized Data Model works as shown below:

The detailed working flow is as follows:

  1. Created and existing Data Stores are verified per AI Agent according to the defined naming conventions.

  2. Table structures are verified to contain all required fields in accordance with default extensible Data Model structures. Simultaneously, the existence of any additional _bi columns is checked.

  3. ETL jobs process the content, obtaining data deltas on schedule from source Data Stores and moving it to Analytics data tables under the dm schema, following the same naming patterns adopted for source tables.

  4. A default ETL schedule is set up to run every 15 minutes.

  5. As the data is transferred and available via dm tables, it is possible to clean up the sources.