Advance Predict Agent

Agent Description:

The Advance Predict Agent is a lightweight, business intelligence solution that processes historical sales data to generate trend-based insights and forward-looking forecasts. It is designed to assist decision-makers by identifying performance patterns, seasonality, and upcoming demand shifts for specific products or regions.

Purpose and Components
  • Purpose: To extract structured sales data from historical documents and provide short-term sales predictions and performance insights.
  • Components:
    • A data reader that parses historical sales from PDFs.
    • An insight generator that detects trends and performs 7-day predictive forecasting.
Supported Capabilities
  • Reading and converting historical sales data from scanned PDFs or digital documents.
  • Structuring sales records by date, product, region, and amount.
  • Identifying growth patterns, seasonality, anomalies, and dips in performance.
  • Forecasting short-term (7-day) sales based on recent trends.
  • Presenting clear summaries for business planning and inventory preparation.
LLM Used
  • Google Vertex
    Note: To learn more about the LLM and to modify its behavior, refer to the Configuring LLM settings section.

Sub-Agents

1. Historical Data Reader

  • Role: Data Reader.
  • Scope: Ingest and clean raw sales data from PDF files.
  • Description: This subagent receives a PDF from the historical sales folder and uses a PDF Reader Tool to extract all tabular records. It parses the sales data into a clean structure with the following fields:
    • Date
    • Product Name
    • Region
    • Sales Amount

    The data is sorted by date, cleaned for accuracy, and de-duplicated before passing it to the next subagent in JSON format.

  • Tool Used:PDF Reader Tool

2. Sales Insight Agent

  • Role: Insight Generator.
  • Scope: To detect performance trends and forecast future sales volume.
  • Description: This subagent takes the structured sales data in JSON format and performs a detailed analysis to detect trends, such as daily or weekly growth, seasonality, and unexpected sales spikes or drops. It then applies lightweight forecasting techniques (e.g., moving averages) to predict expected sales for the next 7 days per product. The output includes:
    • Clear trend summaries
    • Short-term sales projections
    • Flags for anomalies or sales fluctuations
  • Tool Used: No additional tool (uses internal logic and LLM reasoning)
Tools Used
  • PDF Reader Tool: for extracting historical tables from scanned or digital PDF documents.
Note: For details on modifying the Tools, refer Tools Library section.
Agent Workflow Behavior Summary
  • The Historical Data Reader loads and parses past sales records from PDF documents, returning structured and chronological data.
  • The Sales Insight Agent analyzes this data to find trends and predict 7-day forward sales based on current momentum and patterns.
  • The result is a structured summary containing trend insights, sales forecasts, and any anomaly detections to inform business planning.
Sample Questions:
  • Give me data for next 7 days
  • What is the expected revenue for the next 2 weeks?
  • How has our average daily revenue changed over time?