Studio: Getting Started - Basic Settings – Blip | Blip Help

Introduction
Creating and configuring agents in Studio
2.1 How to create an Agent block
2.2 How to configure the Agent block
2.2.1 Renaming the Agent
2.2.2 Configuring the Instructions tab
2.2.3 Configuring the Knowledge Base
2.2.4 Configuring the Exit Conditions tab
2.2.5 Configuring the Tools tab
How do you put your intelligent contact into production?
Complete Orchestrator example
FAQ AI Agents example

Introduction

Studio was designed to allow you to create and customize Artificial Intelligence agents visually and intuitively. With it, you can customize your agent's behavior, define its knowledge base, establish exit conditions for the flow, and integrate external tools.

This tool centralizes all the resources needed for you to intuitively build and manage intelligent agents.

Creating and configuring agents in Studio

Note: At the end of this article, you will find the JSON files for the configuration. You can copy them and import them directly into Studio to make any adjustments you want.

How to create an Agent block

On the Studio screen, click Add block and choose the Agent option.

A new Agent block will be added to Studio.

How to configure the Agent block

By clicking the created Agent block, you can configure it through the side menu.

Renaming the Agent

First, you can rename the block to make it easier to identify. In the following example, the Intelligent Contact will be an FAQ about the Agent.

Configuring the Instructions tab

In this section, you can customize your Agent to meet your needs. Choose the language model (LLM), set the temperature and maximum number of tokens, enter the agent's initial instructions, and more.

Precise configuration is essential for optimizing your agent's performance.

To configure a model, click the Configure agent button next to the model indicator on the Instructions tab. The Configure Instructions screen will be displayed.

On the Model tab, you can choose the version of the model to be used from the drop-down menu.

Studio now supports multiple language models, increasing flexibility when creating agents and automations that are better suited to different use cases.

This expansion allows you to choose exactly the model that best meets your needs for file interpretation, performance, and the desired depth of response for each conversational flow.

See the available models in our AI Model Selection Guide.

To expand your agent's Instructions (Prompt) field in Blip Studio, follow the steps below:

1. Open the Agent panel: In the flow, click the Agent box you want to edit to open the settings side panel on the right.

2. Access expansion mode: In the upper corner of the Instructions field, hover over the text block or click the Expand icon (two diagonal arrows or the maximize button) in the upper-right corner of the text box.

3. Edit the prompt in full-screen mode: The field will open in a larger, centered window (modal), allowing you to easily view and edit the entire prompt text.

Checking available AI models

To check and compare all Artificial Intelligence (LLMs) models available in Blip Studio, follow these steps:

1. Open the Agent settings: In your flow panel, click the desired agent and then the Configure agent button (or the gear icon).

2. Access model selection: On the Model tab, locate the Select model field, where the current model is listed.

3. Open the comparison screen: Click the Compare models button, located directly below the parameters at the bottom of the configuration column.

4. Browse the available options: A modal window (Compare and choose an LLM model) will open, displaying the complete list of models organized by category (OpenAI, Gemini, Open Source, Cost-effective, etc.).

Configuring the AI Model and Reasoning Parameters

To access these settings, follow the steps below:

Click the box with the Agent's name.
Access the “Instructions” tab.
Click the "Configure Agent" button.
Then select the "Model" tab, where you will find the reasoning parameters.

1. Selecting the AI Model

The first step is to choose the technology that will power your agent, which can be models from OpenAI or Google Gemini.

The Studio interface will automatically adjust to display or hide specific parameters depending on the restrictions of the selected technology.
Unlike traditional linear chat models (such as GPT-4.1), reasoning models have an internal preliminary processing structure that acts as a chain of thought before delivering the final result.

2. Configuring Model Parameters

The parameters allow you to adjust how the agent thinks, the response length, and the tone of voice.

Reasoning Effort

This parameter controls the depth of the model's internal reasoning (the hidden “chain of thought” step) before generating the final response.

The higher the configured reasoning effort, the greater the token consumption and response time tend to be.
Higher levels result in higher-quality output.
Attention: The options available on the screen change dynamically according to the selected model. Some examples include:
- None: Disables preliminary reasoning for a direct response; available for models such as gpt-5.4, gpt-5.6-luna, and the Gemini Flash family. The o4-mini model, for example, does not accept this option because it always reasons by design.
- Minimal: Minimal reasoning, available in models such as gpt-5, gpt-5.1, and Gemini families.
- Low, Medium, and High: Levels proportional to complexity, supported by all reasoning model families.
- Very High and Maximum: Extra-effort levels exclusive to specific top OpenAI models, such as gpt-5.6-luna. These options are not mapped to Gemini.

Response Style

This parameter defines whether the agent's response should be more concise and direct or more elaborate.

It does not control internal reasoning, but rather the density and completeness of the final response delivered to the user.
You can choose options such as a concise and direct response (Precise), an intermediate level (Balanced), and a complete and detailed response (Creative).
Important rule: The “Custom” option applies exclusively to Google Gemini reasoning models, as does “Custom Temperature.”
- Temperature is a numerical value that controls the AI's randomness or creativity:
  - Low temperatures generate predictable, direct, and factual responses.
  - High temperatures allow for greater creativity, variation in
    words, and originality.

Token Limit

Defines the approximate maximum number of tokens that can be used in the agent's response. If the response requires more tokens than the defined value, it will be interrupted halfway through.

Attention for Reasoning Models: These models naturally consume more tokens due to their preliminary internal processing. If the configured limit is too low, the AI may consume much of the token quota just “reasoning,” causing the final response to be truncated (incomplete) or even empty. Therefore, always consider setting higher token limits when using reasoning models.

Where to access:
Configure Agent → Model → Available LLM Models

In this same tab, you can also:

Set the model temperature;
Set the maximum number of tokens;
Enable/disable message history (short-term memory).

On the Response tab, you can:

Define whether the agent's response will be sent to the user through the service channel or stored in a variable. When stored in a variable, it can be used by other agents.

Define the agent's response format. You can choose from several formats, including a custom format.

On the Interpretation tab, you can:

Define the types of files the agent can interpret, as well as configure how to handle the case where the agent receives an unsupported type.

After defining the agent settings, simply click Save. If you have not made any changes, click Cancel.

Still on the Instructions tab, you can configure instructions for the Agent. For each instruction, you can select its level: System, User, Agent, History. Let's go through each one:

System Level

The system level defines the AI agent's overall behavior, persona, and restrictions. It is the most fundamental instruction, establishing the context and rules for how the model should respond, regardless of the user's specific input.

Usage example: "You are a Python programming expert. Respond clearly and objectively, providing code examples whenever possible."

User Level

The user level represents the end user's direct input or query. It contains the specific question or instruction that the user wants the AI to solve in a given interaction.

Usage example: "What is the capital of France?"

Agent Level

The agent level represents the responses generated by the AI agent itself. In a multi-turn conversation, the AI's previous messages can be included as context for subsequent interactions, maintaining dialogue consistency.

Usage example: "The capital of France is Paris."

History Level

The history level allows you to use message histories from other agents. This makes it possible to integrate their context into the current agent.

Usage example:

In practice:

Add an instruction

Select the instruction level:

Enter the instruction:

In the History instruction, you will see the Manage history icon. It allows you to configure how message histories from other agents should be inserted. You can configure the number and order of messages.

In the System and Agent instructions, you will see a shield icon. It provides suggestions for guardrails that you can add to your agent. Simply copy them and insert them into the instruction.

For this example, we will create a new system instruction and add a guardrail for the knowledge base.

For each instruction, you can remove or duplicate it using the buttons:

You can move an instruction by clicking and dragging it using the side button (six dots) that appears when you hover over it. This allows you to organize the order of the instructions, which is crucial to the agent's behavior.

If the context history option is enabled in the model settings, it will also appear on the Instructions screen and can be ordered with the others.

Configuring the Knowledge Base

Moving on to the Tools tab, this is what we see when opening it for the first time:

For the agent to access information, it needs catalogs. These catalogs centralize knowledge from files and URLs, organizing everything related to the same topic. To learn more about catalogs, access the tutorial on the Knowledge Base.

Since your agent was just created, it does not yet have any linked catalogs. To add one, simply click the "Tools" "+ Add Tools" + “Knowledge Base” button.

When you click Knowledge Base, the Add Catalogs window opens. To create a new catalog, click "Create catalog". You will be redirected to the Knowledge Base management screen, where you can create or remove catalogs.

Following our FAQ about AI Agents example, we will create a catalog using an XLSX file with questions and answers and a URL, containing information about the types of files accepted in an AI Agent's knowledge base.

After creation, the catalog will be listed on the Add Catalogs screen in Studio and will already be available to link to your agent. For more details about creating catalogs, see the Knowledge Base tutorial.

To link a catalog to the agent, select the corresponding checkbox. After selecting it, the catalog will be available for queries.

Additionally, you can manage your files and URLs in the knowledge base. Simply select the desired items, such as XLSX files or links, to define which information the agent should use.

Running the Intelligent Contact

Now is a good time to test the Intelligent Contact (IC) with the Agent. To do this, connect the Start block to the Faq AI Agent block and publish the IC flow. You can interact with the agent in the flow test chat.

Configuring the Exit Conditions tab

To configure the Exit Conditions tab, we will create a new flow. The idea is to build a Blip Intelligent Contact with three agents:

Plans Agent: Specialized in the plans offered by Blip.
Products Agent: Specialized in available products.
Orchestrator Agent: Responsible for understanding the user's intent and directing them to the correct specialist agent.

In this scenario, we will also add deterministic blocks to show that they can be combined with AI Agent blocks.

To get started, we will add the deterministic blocks that will welcome the user and collect their name. To do this, we will copy ready-made blocks from the Block Library and adapt them for our example.

The initial flow looks like this:

Now you can create the orchestrator agent and define its instructions.

In this example, it will have three system instructions:

The first defines its role as an orchestrator.
The second instructs it to address the user by name, using a variable for personalization.
The third is a guardrail to ensure that the orchestrator does not send unnecessary messages to the user.

The knowledge bases for the specialist agents will be the URLs for Blip's products and plans.

The instructions for each agent will be as follows:

Role Definition: Describes the agent's area of expertise.
Competition Guardrails: Prevents the agent from mentioning competitors or similar services.
Tone-of-Voice Guardrails: Ensures that responses are in Portuguese and maintain a friendly tone.
Grounding Guardrails: Ensures that all information provided is based exclusively on the knowledge base.
User's name: Ensures that the agent addresses the user by name.

Now simply create the agents, configuring them with these instructions and the knowledge base, as shown in the previous steps.

Blip plans specialist agent:

Blip products specialist agent:

Now you can configure the exit conditions. They guide the agent regarding the next block to which the user will be directed, ensuring the flow continues correctly.

Let's start with the Orchestrator. On the Exit Conditions tab, click the + Add routing button.

Under Definitions, enter the name, description, and destination block, then save the instruction. Create an exit condition for each specialist agent.

You can also define an exit condition for exception cases.

Now, for each specialist agent, we will create an exit condition to return to the orchestrator agent. This allows the user to return and change context.

This is what the flow looks like:

Now all that's left is to test it!

Configuring the Tools tab

The Tools tab allows you to configure instructions for the agent to interact with external resources. To see a practical example, let's return to the Orchestrator agent.

When you click Add tool, a list of options will be displayed. For each one, in addition to defining a name, you must provide a description—which guides the agent on when to trigger the action—and configure the specific data needed for its execution.

Let's use the Set contact tool as an example, which is used to store user data. Since we have already captured the name, we will use it to fill in the contact details. To do this, simply change the name, enter the description, and map the {{nome}} variable to the corresponding field.

You can add conditions for setting the contact. For example, you can include a validation for the user's name.

To test how it works, we will adjust the instruction we used to pass the user's name. Instead of using the {{nome}} variable, we will change the instruction to retrieve the name directly from {{contact.name}} and validate that the agent can still address the user by name:

How do you put your intelligent contact into production?

To put your intelligent contact flow into production (publish it) in Blip Studio, simply click the blue "Publish flow" button in the upper-right corner of the screen.

Step-by-step publishing instructions:

Click the "Publish flow" button in the upper-right corner.
If the system displays a screen confirming changes or warnings, read the details and confirm the action.
The interface will indicate that publishing was completed successfully.

How to check whether it worked:

After publishing, test it by sending a message to the connected channel, such as the test WhatsApp or your bot's active channel, and check whether the response follows the structure of the newly published flow.

Complete Orchestrator example

Download the example here.

FAQ AI Agents example