Agents
An agent is a workflow where a language model decides what to do next at each step. It receives a task, reasons about how to approach it, and calls tools to complete the task.
Typical use cases include:
- Answering multi-step questions that require data lookups
- Routing tasks to different workflows based on the input
- Automating decisions that depend on intermediate results
- Building interactive assistants with a chat interface
Create and prompt an agent
An agent workflow connects a model, a prompt, and tools to an Agent Prompter node. The Agent Prompter runs a reasoning loop: the model reads the task, decides whether to call a tool, reads the result, and repeats until it has an answer.
- Create an agent: Set up an agent workflow with a model and a system prompt.
- Prompt an agent: Start the reasoning loop using the Agent Prompter or the Agent Chat View.
Add tools
Tools are KNIME workflows that an agent can call during its reasoning loop. The agent selects which tool to use based on the tool's description.
- Turn a workflow into a tool: Prepare a workflow so an agent can call it.
- Provide tools: Register tools and make them available to an agent.
- Pass data to tools: Let tools read and process data tables during execution.
Extract conversation data
- Extract conversation data: Retrieve the full conversation history from an agent chat session as a table.
Tutorial
Follow a complete example that combines the concepts above to build an agent: