Cloudera Docs

Creating an agent manually

Create custom AI agents manually in Agent Studio within Cloudera AI to define specific roles, goals, and LLM models for your workflows.

  1. In the Cloudera console, click the Cloudera AI tile.
    The Cloudera AI Home page is displayed.
  2. Click on the name of the workbench.
    The Cloudera AI Workbench Home page is displayed.
  3. In the left navigation pane, click AI Studios.
    The AI Studios page is displayed.
  4. On the Agent Studio tile, click Open.
    The Agent Studio page is displayed.
  5. Click on Create.
    The Create New Workflow window is displayed.
  6. Enter a Workflow Name.
  7. Click on Create Workflow.
    A draft workflow is created and a Edit Workflow window is displayed.
  8. Click on Create or Edit Agents.
    The Create or Edit Agent window is displayed.
    Figure 1. Create or Edit Agent window
  9. Specify the following details:
    1. Name – Name of the agent.
    2. Role – Description of the role this agent plays in the workflow.
    3. Backstory – Context of history of the agent.
    4. Goal – The primary objective or purpose of the agent.
    5. LLM Model – LLM model that the agent will use in the workflow.
    6. Agent Icon – Upload a file for the agent icon.
    7. Add Tools – Create custom tools for an agentic workflow during the workflow creation process to extend platform functionality. For more information, see . Creating custom tools.
      To design an effective Agent, you must enable interaction and the creation of concrete, shareable results that are artifacts. Cloudera recommends integrating specific tools that allow the agent to create Artifact Files.
      Use the following recommended tools:
      Table 1. Recommended tools and its function
      Tool name Primary function
      Artifact Files Read/Write Tool Enables the Agent to ingest, process, and persist various file types as workflow output.
      Code Execution Tool Allows the Agent to run embedded or generated code snippets, to produce data transformations, statistical analysis, or complex computational results.
    8. Add MCP Servers – MCP servers can be incorporated during the workflow creation or modification process, to enhance the functionality of agents within a workflow. For more information, see Adding MCP servers to agents
  10. Click on Create Agent.
    The newly constructed agent is displayed and the following workflow displays:
    • A conversational task.
    • The conversational task is assigned to the Cloudera Manufacturing industry Analyst agent.
    • The workflow incorporates the two added tools.
    Figure 2. Agent in a workflow