Cloudera Docs

Key features of Agent Studio

Cloudera AI Agent Studio capabilities include low-code workflow orchestration, custom tool extension, model integration, and Phoenix observability for production-ready AI agents.

  • Design agentic workflows with a low-code interface –– Agent Studio leverages workflows as the foundational structure for AI agents. A workflow represents a network of collaborative agents working together to perform a sequence of interconnected tasks. Use the low-code interface to perform the following actions:
    • Create agents and assign them specific tasks and tools.
    • Define agentic workflows as either conversational or task-oriented.
    • Assign a manager agent to oversee and coordinate interactions between agents.
  • Leverage pre-built and custom workflow templates –– Agent Studio includes a comprehensive library of pre-built workflows and tools to accelerate development. Additionally, you can create custom workflows and tools, which can be saved and managed as reusable templates across the team.
  • Configure agents and tasks –– Agents are intelligent entities capable of decision-making, running actions, and collaboration. Configure your agents within the UI to perform the following actions:
    • Define each agent's role, goal, and contextual backstory.
    • Assign specific tasks and tools to individual agents.
    • Use AI-assisted authoring to generate agents using natural language input.
  • Develop and extend custom tools — Tools allow agents to interact with external systems, such as APIs, databases, or business logic, beyond the native scope of LLMs. Expand your agent capabilities through the following development paths:
    • Deploy built-in tools or develop custom tools.
    • Extend existing tools by modifying underlying logic, parameters, or API integrations.
    • Develop new integrations directly within Cloudera AI Workbench notebooks and Python sessions
  • Integrate workflows with AI models –– Integrate your workflow with the LLM of your choice by registering a model and configuring the required API credentials.. You can connect to Cloudera AI Inference service or use any OpenAI-compatible model provider.
  • Test and debug workflows –– Agent Studio ensures workflow quality and correctness through comprehensive testing and debugging features. Optimize your configurations by performing the following actions:
    • Run workflows with test inputs and visualize their execution paths using interactive flow diagrams.
    • Access step-by-step logs and animations in real-time.
    • Use playback mode to analyze and debug specific workflow steps.
  • Deploy workflows as endpoints and applications –– Transition finalized agent configurations into production environments immediately. Leverage the following deployment capabilities:
    • Automatic generation of a Workbench Model endpoint upon publication.
    • Creation of a default user-facing application for immediate interaction with the liveworkflow.
  • Monitor workflow performance –– Agent Studio integrates natively with the Arize Phoenix, open-source observability tool. Use this integrated tooling to perform the following actions:
    • Monitor real-time operational latency, token consumption, inputs, outputs, and execution traces.
    • Access to a dedicated observability panel within the Workflow UI and or view Phoenix dashboards in Cloudera AI Workbench.
  • Build User-facing applications and custom UIs – Each deployed workflow includes a default, functional front-end interface for users. To customize the consumer experience further, build your own applications using the Agent Studio Python SDK and APIs.