Building a Company Financial Report

This demonstration covers a simplified (educational) example of monitoring the latest events associated with a publicly traded company. An API connector is used to collect data from a specific company (recent news, management team, and financials), which is then semantically analyzed using LLMs with results post-processed and outputed into a report.

One-Shot Reports

This demonstration showcases the concept of a one-shot report, where a generative AI expert agent performs various analytical tasks that are aggregated into a single report. Many analytical tasks aim to produce a document that builds an evidence/data-driven argument to support downstream decision-making.

Lunar integrates this concept into its architecture, allowing the end-to-end organization of the analytical cycle from data source collection to communication of results. While some reports can be fully automated, Lunar users commonly treat one-shot reports as a first draft, which concentrates all relevant evidence into a single point of access and performs preliminary analytical processes.

Workflow

In this example workflow, the Yahoo Finance API is used to collect basic company data for a specified company. Given the stock identifier as an input parameter, the workflow starts by using an external finance API to collect fundamental, shareholder, management team and event data for a target company. All data is then semantically analyzed using LLMs, post-processed, and all results are output into a report.

Contributors

This workflow was developed by Alex Bogatu and adapted by Danilo Miranda and Pedro Rocha.