Agentic Workflow
An AI agent uses artificial intelligence to perceive its environment, make decisions, and take actions to achieve specific goals.
More technically, an AI agent in our context is an entity based on an LLM model and equipped with a specific set of tools (MCP tools).
An agentic workflow involves multiple AI agents working together towards achieving a common objective.
- Go to
Agentic Workflow
and click onCreate new workflow
button. The window already contains a pre-populated test code example:
This Python script demonstrates the creation of an agentic workflow named weather_time_agent using the Google Agent Development Kit (ADK) and the gemini/gemini-2.0-flash model through LiteLLM. The agent is designed to answer user questions about the current time and weather in a specified city.
-
Before executing the agentic workflow, set the credentials for LLM provider, e.g.:
-
Click the 'Run' button. If the code compiles correctly, the 'Agent Interaction' tab will open. In this tab, you can interact with the agent:
-
Click on 'Save' to save the agentic workflow for further use.