vFunction Query Engine with GenAI Integration
Overview
The vFunction Query Engine bridges the gap between natural language and technical data. By translating user written prompts into internal vFunction DB queries, the Query Engine opens new ways for users to explore and retrieve information based on their specific needs. This offers a new level of accessibility, empowering teams to uncover application specific insights for their app requirements. The Query Engine was introduced in Server v4.2 and later.
The vFunction Query Engine leverages public Large Language Model (LLM) integrations to translate natural language queries into corresponding SQL statements that can be executed on the vFunction database. This process involves providing the LLM with the natural language input, a SQL query template and the vFunction database schema. Importantly, no table data or application-specific values are shared with the LLM, ensuring that no client or application data is exposed to it. Additionally, the queries are run without any write privileges so the data remains secure and uncompromised.
Access the vFunction Query Engine
Take the following steps to access the vFunction Query Engine:
- Log into the vFunction Server UI
- Select an Application by clicking the dropdown arrow in the top-left corner of the UI
- Navigate to the Analysis tab in the top-center menu bar
- Click Query Engine below the Actions Header in the left-hand sidebar
- Type your natural language query into the Your query text field of the Query Engine dialog box
- Click the blue Submit Query button and wait for the results to be populated
Creating Queries for the vFunction Query Engine
Natural Language Queries can be created for an entire Measurement or for contextual questions related to a specific Node in a Call Tree or to a specific Domain. Once the query is executed successfully, users are presented with the results, along with an option to download the data as a CSV.
Only the query is sent to the GenAI provider. The result of the query is in a form that the vFunction database can process. Only then is the query executed on the data stored within the vFunction Server and presented to the user.