Overview: Build Your First Workflow

Lesson Overview

In this video, you will learn how to build your first workflow in AirOps to analyze Google search results for a specific keyword. By the end, you will be able to input a keyword and receive an analysis of the content performing well on the search results page.

• 0:00: Introduction to building your first workflow
• 0:25: Creating an input for the search keyword
• 0:37: Fetching the first page of Google search results
• 1:27: Using an LM (Language Model) step to summarize the results
• 2:48: Publishing the workflow and introducing Grids

Key Concepts

Inputs Node

Every workflow in AirOps has an inputs node where you define the inputs for the workflow. In this example, a single input called "keyword" is created to represent the search keyword to be analyzed.

Google Search Step

The Google Search step in AirOps allows you to fetch the first page of Google search results for a given keyword. You can pass the keyword input into the search query to dynamically fetch results based on the input.

LM (Language Model) Step

The LM step in AirOps enables you to use a language model to analyze and summarize data. In this example, the LM step is used to analyze the Google search results and provide insights on the types of content performing well for the given keyword.

• Prompts can be set up with a system message to define the model's role and objective
• Chat messages can be used to provide specific instructions or requests to the model
• Output from previous steps can be easily incorporated into the current step's prompt using the variable selector

Grids

Grids in AirOps allow you to run workflows at scale. You can input multiple values in the first column and run the workflow for each value. The results are displayed in the output column, and you can review the model's response for each input.

• Grids provide a powerful way to leverage a single workflow for multiple inputs
• Results can be downloaded as a CSV for further analysis
• Additional workflows can be chained to perform more complex operations

Key Takeaways

1. AirOps enables users to build workflows that fetch and analyze data from various sources, such as Google search results.
2. Inputs can be defined for a workflow, allowing dynamic data to be passed into steps like the Google Search step.
3. Language Model (LM) steps can be used to analyze and summarize data, with prompts set up to guide the model's output.
4. Grids provide a scalable way to run workflows with multiple inputs, displaying results in an easy-to-review format.
5. Workflows can be published and shared, and grids can be used to chain multiple workflows together for more complex operations.

‍