Configure the AI Agent Node in Agent Studio
About the AI Agent Node
The AI Agent Node is the central component in Agent Studio that gives your flow-based agents their intelligence. It processes incoming information, interprets instructions, and decides what action to take next, such as generating a reply, using a tool, or passing data along the flow. Unlike a standard node with fixed logic, the AI Agent Node can adapt its behavior based on the context of the conversation and the specific instructions you provide.
Important Note: Agent Studio is designed for maintaining existing flow-based agents. For creating new agents, please use the current, recommended agent-building experience available in your account.
Core Configuration Elements
Configuring an effective AI Agent Node involves setting up several key elements that work together to define the agent's behavior.
Prompt
The prompt contains the instructions that guide the AI agent. A well-defined prompt tells the agent its role, goals, limitations, and when to use available tools. You can make prompts dynamic by inserting variables from your flow.
- Variable Sources: You can pull information into your prompt from various sources, including the conversation history, data captured earlier in the flow, or details from a connected contact record in your CRM.
- Prompt Enhancement: The system can automatically refine your instructions to improve clarity and performance before they are sent to the AI model.
Model
This setting selects the underlying AI language model that powers the node's reasoning. Different models offer varying balances of capability, speed, and cost. Choose a model that aligns with the complexity of the task your agent needs to perform.
Mode
The mode determines the node's primary function.
- Conversational Mode: The agent interacts directly with a user, generating responses and managing a dialogue.
- Task-Based Mode: The agent works silently in the background to execute actions, such as extracting data or making decisions, without producing user-facing messages.
Tools
Tools extend what your agent can do beyond just generating text. Based on the context, the agent can decide to use a tool to retrieve information, route the flow, or trigger an external action.
Available tools include:
- Router
- Knowledge Base Search
- Search the Web
- MCP Server Actions
- API Call
- Content Generation (Text, Image, Video, Audio)
Variables
Variables allow the agent to work with data. You can pass information into the node as input variables to provide context. The agent can also create new runtime variables to capture structured information (like a name or an email address) from an unstructured conversation, which can then be used later in the workflow.
How to Configure the AI Agent Node
Step 1: Add the Node to Your Flow
Drag and drop the AI Agent Node from the node library into your Agent Studio flow canvas.
Step 2: Define the Agent's Prompt
In the node's settings, write clear instructions in the prompt field. Define the agent's role, its task, the desired tone, what it should avoid, and when it should use the tools you will attach.
Step 3: Select an AI Model
Choose an AI model from the dropdown menu that best suits the needs of this particular agent.
Step 4: Choose the Operating Mode
Select either Conversational Mode or Task-Based Mode based on whether the agent should talk to users or work silently.
Step 5: Attach Tools
In the tools section, select which capabilities you want to make available to this agent. The agent will reference its prompt and the conversation context to decide if and when to use them.
Step 6: Define Runtime Variables
Specify what key pieces of information you want the agent to try and capture from the interaction. For example, you can instruct it to capture a customer issue or a demo request date and save them as new variables for the flow.
Step 7: Connect the Node in the Flow
Connect the output of a previous node (like a Start Trigger) to the input of your AI Agent Node. Then, connect the output of your AI Agent Node to the next step in your workflow. The node will execute when it receives data from an incoming connection.
When to Use the AI Agent Node
Use this node in scenarios where you need an agent to handle unpredictable inputs, manage multi-turn conversations, make context-dependent decisions, or combine reasoning with action using tools. It is ideal for building intelligent customer support qualifiers, interactive FAQ handlers, or dynamic data capture assistants.