To build a lead classification automation in VectorShift, we'll follow these steps:

1. Create a Pipeline for Classifying a Single Company:

   • Use an Input Node to accept the company name.
   • Perform a Google Search on the company name using the Web Search Node.
   • Pass the company name and the search result snippets to an OpenAI Language Model.
   • Prompt the Language Model to classify the company into one of three categories: automobiles, consumer packaged goods, or retail. If the company doesn't fit into these categories, the model should respond with "other".
   • Send the classification result through an Output Node.
   • Name this pipeline "Classify Leads" and deploy it.

2. Create a Pipeline to Automate the Classification Process:

   • Use a Read from Sheet Node to read the list of companies from a Google Sheet.
   • Call the "Classify Leads" pipeline using the Pipeline Node, passing in the list of companies.
   • Enable the "List Mode" option in the Pipeline Node to run the classification on each company in the list.
   • Use a Write to Sheet Node to write the classification results back to the Google Sheet.

By following these steps, you'll have a fully automated lead classification system that can handle a large number of companies efficiently.

To classify a single company, we will build a pipeline that performs the following steps:

1. Take the company name as input.
2. Perform a Google search on the company name and extract the top 5 search snippets.
3. Pass the company name and search snippets to an OpenAI language model, which will classify the company into one of three categories: automobiles, consumer packaged goods, or retail. If the company doesn't fit into any of these categories, the language model will respond with "other".
4. Output the classified category.

The key components of this pipeline are:

• Input Node: Accepts the company name as input.
• Google Search Node: Performs a Google search on the company name and extracts the top 5 search snippets.
• OpenAI Language Model Node: Uses the company name and search snippets to classify the company into one of the three categories or "other".
• Output Node: Sends the classified category as output.

By building this pipeline, you can easily classify a single company by providing its name as input, and the pipeline will output the corresponding industry category.

To create a pipeline to classify a list of companies, we'll follow these steps:

1. Build a Pipeline to Classify One Company:

   • Use an Input Node to accept the company name.
   • Perform a Google Search on the company name using the Web Search Node.
   • Pass the company name and the search result snippets to an OpenAI Language Model.
   • Prompt the Language Model to classify the company into one of three categories: automobiles, consumer packaged goods, or retail. If the company doesn't fit into these categories, it should respond with "other".
   • Send the classification result through an Output Node.
   • Name this pipeline "Classify Leads" and deploy it.

2. Create a Pipeline to Classify a List of Companies:

   • Use a Read from Sheet Node to read the list of companies from a Google Sheet.
   • Call the "Classify Leads" pipeline using the Pipeline Node, passing in the list of companies.
   • Turn on "List Mode" in the Pipeline Node to apply the classification to each company in the list.
   • Use a Column List Writer Node to write the classifications back to the Google Sheet.

By following these steps, you can create a pipeline that can classify a list of companies into the specified categories, automating the repetitive task of classifying each company.

The key takeaways from this tutorial are:

1. Break down repetitive tasks into their atomic units - in this case, classifying a single company into an industry category.
2. Build a pipeline that can perform this atomic task, using an input node, a Google search, and an OpenAI language model to classify the company.
3. Create a second pipeline that can read data from a Google Sheet, call the classification pipeline on each item in the list, and write the results back to the sheet.
4. Leverage Vector Shift's list mode to efficiently apply the classification pipeline to every item in the list.

By following this approach, you can automate repetitive tasks at scale, saving time and effort. The ability to build modular pipelines and reuse them across different datasets makes Vector Shift a powerful tool for automating a wide range of business processes.