FoundationaLLM enables the creation of analytic copilots that provide users with support for self-service analytics. In this example, an agent is configured to answer questions over sales data sourced from an enterprises sales database. When the user asks a question, the large language model capability to generate SQL and code is used to first identity the data requested, then query and join the data if coming from multiple tables and then compute the desired result. The querying and calculation is done without requiring the user to write any SQL or any code, just asking for the data in plain English. With the data in hand, the agent can be further prompted to use a charting tool to produce a bar chart and provide the inline with its response to the user.