Skip to main content
This guide describes the complete process for creating a new AI agent using the creation wizard. The flow is divided into sequential steps, allowing you to configure the name, type, role, behavior, and AI model that will be used by the agent.

​
Creating a New Agent

You can create a new agent using one of the available New Agent buttons. New Agent button The New Agent button is available in two locations:
  • In the upper right corner of the screen
  • In the center of the empty state when no agent exists
Clicking this button starts the agent creation wizard.

​
Starting Agent Creation

To start creating an agent, access the Agents section and click New Agent. This will open the creation wizard, which guides the process in steps. Agent creation wizard start In this first step, you must define the agent’s name.

​
Agent name

The What is your agent’s name? field defines the name that will be used to identify the agent within the workspace.
  • The name is required
  • It will be displayed in the agent list
  • It can be changed later in advanced settings
After filling in the name, click Continue to proceed.

​
Selecting the Agent Type

In the next step, you must choose the agent type that best fits your use case. Agent type selection The available types determine how the agent will operate:
  • LLM (Language Model): Conversational agent based on language models (most common)
  • Task: Agent focused on automating specific tasks
  • Sequential: Executes agents in sequence
  • Parallel: Executes agents in parallel
  • Loop: Executes agents in a continuous loop

​
LLM Agent Type

For conversational agents, select the LLM (Language Model) option. LLM agent type selected This type is the most commonly used and recommended for virtual assistants, customer service, technical support, and sales agents. After selecting the type, click Continue.

​
Defining Agent Role and Goal

In this step, you define the agent’s role and main objective. Agent role and goal definition

​
Agent Role

The Agent Role field describes the agent’s function, that is, what role it plays (for example: support assistant, salesperson, analyst).

​
Main Goal

The Main Goal field defines the agent’s main objective, describing what it should achieve during interactions. These fields help guide the agent’s behavior, but can be adjusted later. This step is optional and can be skipped by clicking Skip.

​
Defining Agent Instructions

The instructions step defines how the agent should behave and respond. Agent behavior instructions In the What does your agent do? field, describe:
  • The expected behavior of the agent
  • The response tone (formal, friendly, technical, etc.)
  • Limits and responsibilities
  • Best practices for service or execution
The more detailed the instructions, the more consistent the agent’s behavior will be.
You can use the Review Prompt or Generate with AI buttons to review or automatically generate suggestions.

​
Selecting the AI Model

In this step, you define which AI model will be used by the agent. AI model selection and API Keys management Before selecting a model, you need to configure an API Key.

​
Managing API Keys

Click Manage to open the API key manager. Manage API keys If no key is configured, you can click Add First Key or New Key. Add new API key Fill in the fields:
  • Name: Identifier name for the key
  • Provider: Model provider (e.g., OpenAI)
  • Key: API key provided by the provider
Click Add to save the key.

​
Choosing the AI Model

After configuring the API Key, you will be able to select the AI model. AI model selection In the Search or select a model field, choose the desired model (for example: GPT-4.1, GPT-4.5 Preview, GPT-4o, among others). The displayed list depends on the provider and the configured API key. Click Continue to finalize creation.

​
Agent Created Successfully

After completing all steps, the agent will be created and ready for use. Agent created successfully On this screen, you can:
  • Adjust advanced settings
  • Test the agent in a chat
  • Add tools
  • Return to the agent list
From here, you can proceed to more advanced settings, such as MCP servers, custom tools, and knowledge bases.

​
Final Considerations

The agent creation process was designed to be quick and intuitive. After initial creation, you can refine settings, add tools, and knowledge bases to expand the agent’s capabilities.