Cloudera Docs

Use case: Sequential Idea Generation Workflow

This example illustrates

  1. In the Cloudera console, click the Cloudera AI tile.

    The Cloudera AI Workbenches page displays.

  2. Click on the name of the workbench.

    The workbenches Home page displays.

  3. Click Projects, and then click New Project to create a new project.

    In the left navigation pane, the new AI Studios option is displayed.

  4. Click AI Studios.
  5. Launch the Agent Studio within a project.
  6. Click Create in the Create Agentic Workflow block to create a new workflow.
    The Create New Workflow page displays.
  7. In Create New Workflow page, in Workflow Name, provide Sequential Idea Generation Workflow as the name for your workflow and click Create Workflow.
  8. In Add Agents page, configure the following:
    1. Under Capabilities, turn off both the Is Conversational and Manager Agent toggle buttons.
    2. Under Agents, click Create or Edit Agents. The Create or Edit Agent window displays.
      1. Create the first Agent with the following details:
        1. In Name, specify Web Search Agent as the name of the agent.
        2. In Role, specify Web Search Specialist as the role of the agent.
        3. In Backstory, provide the following backstory You are a reliable web search specialist who has a knack for finding great information online using a web search tool. You are an expert in creating comprehensive search phrases that yield the best search results.
        4. In Goal, provide the following goal Return a summarized set of bullets that cover the most relevant details for the search required.
        5. Click Create or Edit Tools. The Create or Edit Tools page displays.
          1. Select the built-in Web Search Tool.
          2. Click Create Tool from Template.
          3. Click Save Tool and then click Close. The Web Search Tool is now associated with the Web Search Agent and displayed in th Create or Edit Agent window.
      2. Click Create Agent to create the agent.
      3. Click Close after saving the agent.
    3. Click Create Agents and create the second Agent with the following details:
      1. In Name, specify Creative Monetization Strategist as the name of the agent.
      2. In Role, specify Creative Monetization Strategist as the role of the agent.
      3. In Backstory, provide the following backstory You are a creative agent who reviews the information provided as an input and then comes up with ways to make money using this information.
      4. In Goal, provide the following goal Provide a list of 5 bullets in markdown using the information you are given. This agent does not need any tools.
      5. Click Create Agent to create the agent.
      6. Click Close after saving the agent.
    4. Once both agents are created, click Save & Next in the Add Agents page to proceed with the next steps.
    5. In the Tasks page, follow the steps outlined below. Since this is not a conversational workflow, you will need to define specific tasks for each agent to accomplish. Ensure that tasks are created in the correct sequence to maintain a logical workflow.
      1. Set the following parameters for the first task:
        1. In Task Description, provide the description as Make sure to research {topic} and find the latest information in the field.
        2. In Expected Output, provide the expected out put as Provide a list of outputs with interesting facts from what you find about {topic}. Keep your answer in the markdown format.
        3. In Select Agent, select Web Search Agent.
        4. Click Add Task to include this task in the workflow.
      2. Set the following parameters for the second task:
        1. In Task Description, provide the description as With the context provided, generate 5 ideas on ways to make money using that information.

        2. In Expected Output, provide the expected out put as 5 bullets formatted in Markdown

        3. In Select Agent, select Creative Monetization Strategist.

        4. Click Add Task to include this task in the workflow.

    6. Click Save & Next after adding both tasks to proceed to the next step in the workflow configuration.
  9. In Configure page, set the following parameters:
    1. In Max New Tokens, specify 1024. This value represents the maximum number of new tokens the agents and manager agent can generate during LLM calls. Note that certain LLM endpoints may impose restrictions on this value.
    2. Set the Temperature to 0.2. The temperature controls the variability or creativity in the agent's responses generated by the LLM. A higher temperature value results in more diverse and creative responses. A lower temperature value results in less varied and more deterministic responses.
    3. Under Tools, in serper_api_key, specify your Serper API key.
    4. Click Save & Next.
  10. In the Test page, you can choose a topic of your choice to test the workflow. As an example, Video Games is selected.
    1. Click Run Workflow to execute a test of the configured workflow.
    2. Once the test is complete, click Save & Next to proceed to the deployment stage.
  11. In Deploy page, you can perform the following :
    • Click Deploy to initiate the deployment of the workflow. The deployment process typically takes 5 to 7 minutes to complete. Once deployed, navigate to the Applications tab in the left navigation pane to view the deployed workflow.
    • Click Save Template to create a reusable template of the workflow. This template can serve as a reference for future use.
    • Click Test to return to the previous page and re-test the workflow, if necessary.