Agentflows

Important: Agents is currently in private beta and is not yet available to all customers.

Build agentflows in Nintex to automate complex processes and manage decision-based logic. For example, you can create an agentflow to analyze data, evaluate conditions, and generate reports or alerts.

Jump to:

Create your agentflow

Create an agentflow in Nintex using the drag-and-drop agent designer. Each agentflow consists of a series of actions that define its behavior. Actions can interact with third-party applications using connections.

Agentflow actions: Add actions to define what you want the agentflow to do. Actions are the steps the agentflow performs to complete a process. If an action is a connector action, you must create a connection so the action can interact with a third-party application.

For more information about creating an agent, see Create an agentflow.

Using agents with LLMs

When building agentflows, you use a Large Language Model (LLM) whenever the agentflow needs to understand natural language, generate responses, reason, or perform intelligence-driven tasks. The LLM then directs the agentflow’s actions, allowing it to handle tasks, make decisions, and communicate effectively. Within the long list of providers, a few are particularly popular and widely used.

  • OpenAI: Balances performance and ease-of-use, and many tools or platforms already integrate it. For more information on the model parameters, see Chat completions.

  • Google Gemini: Provides multimodal capabilities, such as text, image, audio or video, and integration with Google ecosystem. For more information on the model parameters, see Generate content with the Gemini API in Vertex AI.

  • Anthropic Claude: Provides reliable reasoning, compliance, and long-context handling. For more information on the model parameters, see Models overview.

Access the Agentflows page

  • Go to Agents > All agents

The Agentflows page is displayed. For more information, see Agentflows list