How to build a Market Research Agent

Workflow Overview: Market Research Report Generator

1. User Input

• Node: Stock/ETF/Instrument Query

• What it does:
  The user enters a stock ticker, ETF symbol, or the name of a market instrument they want to research (e.g., "PLTR").

2. Data Gathering

A. Web Search

• Node: Web Search

• What it does:
  Searches the web for the latest news and information about the instrument entered by the user.

  • Input: The user’s query (from the input node).

  • Output: A list of recent news articles and web results related to the instrument.

B. Stock Ticker Info

• Node: Stock Ticker Info

• What it does:
  Retrieves detailed information about the stock or instrument using its ticker symbol via the Polygon API.

  • Input: The user’s query (from the input node).

  • Output: Company details such as symbol, short name, long name, website, business summary, and market cap.

3. Data Processing

• Node: Python

• What it does:
  Processes the web search results to format them as a list of clickable links with short summaries.

  • Input: The output from the Web Search node.

  • Output: A markdown-formatted list of news headlines, each linking to the source and including a short summary.

4. Report Formatting

• Node: Market Research Report (Template)

• What it does:
  Combines all the gathered and processed information into a clean, readable market research report.

  • Inputs:

    • The user’s original query.

    • The formatted news list from the Python node.

    • Company/instrument details from the Stock Ticker Info node.

  • Output: A markdown report with sections for latest news and company details.

5. Output to User

• Node: Output

• What it does:
  Displays the final, formatted market research report to the user.

How the Data Flows

1. User Input →
   feeds into both Web Search and Stock Ticker Info nodes.

2. Web Search →
   results are processed by the Python node.

3. Stock Ticker Info →
   details are also available to the Python node (if needed) and the Template node.

4. Python →
   outputs a formatted news list to the Template node.

5. Template →
   combines all information into a report, which is sent to the Output node for display.

What the User Sees

• A single input box to enter a stock ticker or instrument name.

• After running, a well-formatted market research report appears, including:

  • The latest news (as clickable links with summaries).

  • Key company/instrument details (symbol, names, website, business summary, market cap).

  • Data sources are clearly cited.

In summary:
This workflow takes a user’s stock or instrument query, fetches both real-time news and detailed company data, processes and formats the results, and presents a professional market research report—all automatically.