/langchain-company-lookup

Primary LanguagePythonMIT LicenseMIT

LangChain Company Information Retrieval Tool

This Python script prompts the user to input a company name and retrieves information about the company using OpenAI and SerpAPIWrapper.

Requirements

  • Python 3.7 or higher

Additionally, you need to set the following environment variables with your API keys:

  • OPENAI_API_KEY: Your OpenAI API key
  • SERPAPI_API_KEY: Your SerpAPIWrapper API key
export OPENAI_API_KEY=your_openai_api_key
export SERPAPI_API_KEY=your_serpapi_api_key

Getting Started

  1. Clone the project
  2. Install dependencies
pip install -r requirements.txt

Usage

  1. Open a terminal and navigate to the directory where the script is located.
  2. Run the script with the command python main.py.
  3. When prompted, enter the name of the company you want to retrieve information for. If you do not enter anything, the default value is Prematch GmbH as in the Linkedin original post.
  4. The script will then use the LangChain library to retrieve the following information about the company:
    • Description
    • Founders
    • Funding rounds
    • Location
    • Website
  5. The script will print out the retrieved information.

Code Description

The readme.md file provides a brief overview of how to use the script. Here is a more detailed description of what the script does:

  1. The script imports the necessary modules from the langchain package, as well as the os module.
  2. A prompt template is defined using a string. The template includes a placeholder for the company name.
  3. The user is prompted to input a company name using the input function. If the user does not enter anything, the default value is Prematch GmbH.
  4. A PromptTemplate object is created using the template string and the input_variables parameter. The input_variables parameter specifies that the company variable will be used in the template.
  5. The OpenAI API key and the SerpAPIWrapper API key are retrieved from environment variables using the os.environ.get function.
  6. An OpenAI object is created using the retrieved API key, a temperature of 0, and the text-davinci-003 model.
  7. A SerpAPIWrapper object is created using the retrieved API key.
  8. A Tool object is created using the search.run method from the SerpAPIWrapper object, along with a name, description, and function.
  9. An Agent object is created using the initialize_agent function from the langchain.agents module. The Agent object uses the search_tool and llm objects, and specifies that it is an AgentType.SELF_ASK_WITH_SEARCH agent.
  10. The prompt string is formatted with the company variable using the format method, and is printed to the console.
  11. The run method of the Agent object is called with the formatted prompt string as an argument. The run method retrieves the requested information from OpenAI and SerpAPIWrapper and prints it to the console.

Resources

Credit

The idea of the project originated from a Linkedin post by Dries Faems