Investments Review Agent

Investments Review Agent

Demo that uses LlamaAgents in combination with LlamaExtract, LlamaClassify and LlamaSheets to showcase the features of LlamaCloud in the fields of financial documents, specifically for what concerns investments spreadsheets and board updates/management presentations.

Installation and Usage

Clone the repository:

git clone https://github.com/run-llama/investments-review-agent
cd investments-review-agent

Install the application:

uv pip install .

Export the necessary env variables:

export OPENAI_API_KEY="..."
export LLAMA_CLOUD_API_KEY="..."

Test the workflows with the associated CLI interface to make sure everything works:

# test with a PDF presentation
pres-wf data/Board-Deck-Template.pdf
# test with an Excel spreadsheet
sheet-wf data/portfolio.xlsx

Run the server:

llamactl serve

Access the application at `http://localhost:4501/`.

How it works

From the frontend of the application, you can choose whether to upload a presentation or an excel sheet.

If you choose 'PDF Presentation', you can upload either a Management Presentation or a Board Update Deck:

1. The uploaded document will be uploaded to the S3 storage in LlamaCloud
2. The uploaded document will be classified as either a Management Presentation or a Board Update Deck (LlamaClassify)
3. Based on the classification, details will be extracted from the file and sent back to the frontend for rendering

Find an example in [`data/Board-Deck-Template.pdf`](data/Board-Deck-Template.pdf)

If you choose 'Excel Sheet', you can upload a spreadsheet containing details on an investment portoflio:

• The file will be uploaded to LlamaCloud S3 Storage
• It will be parsed by LlamaSheets and the data will be extracted and downloaded as parquet files
• The parquet files will be converted to markdown tables
• An OpenAI model will create a summary of the investment portfolio trends and performances.

Find an example in [`da