Components of Agentic AI Studio
- 13 Mar 2025
- 4 Minutes to read
- Print
- PDF
Components of Agentic AI Studio
- Updated on 13 Mar 2025
- 4 Minutes to read
- Print
- PDF
Article summary
Did you find this summary helpful?
Thank you for your feedback
Admins can effortlessly build and customize powerful AI-driven solutions by easily connecting various components. These components work together to create intelligent, autonomous agents capable of understanding inputs, reasoning through tasks, orchestrating workflows, and producing meaningful outputs.
To ensure a functional Agentic AI Flow, the following mandatory components must be included:
Chat Input
User Details Extractor
Prompt
Orchestration Agent
Tools
Chat Output
Chat Input
A Chat Input is a key component of Agentic AI Studio. It allows users to enter messages that initiate a conversation with the Agentic AI, whether it's for customer support, assistance, or any other service. A sample screenshot is shown below:
Figure: Chat Input
It appears as Text Box where the End User types in their messages. It’s a text-based communication with Agentic AI. Chat Input can be submitted using either pressing Enter or clicking Submit button. It allows a smooth conversational flow between the End User and Agentic AI.
Example
End User’s Query on leave balance such as “How can I get to know my leave balance?”
User Details Extractor
The User Details Extractor component is generic-purpose Input field on the Playground which allows the End User to enter text.
Figure: User Details Extractor
It consists of a simple text field, typically with a button or action trigger such as Submit. Text Input can be single-line or multi-line. It is mainly used for transactional purposes.
Example
Search Queries or Commands
Chat Input vs User Details Extractor
Both Chat Input and User Details Extractor components are important for gathering user input, yet they are designed for different tasks. Let’s us see the difference between Chat Input and Text Input.
Feature | Chat Input | User Details Extractor |
|---|---|---|
Purpose | It used primarily for real time communication | It is primarily used for data entry tasks |
Interaction Type | It operates in real-time chat interactions with Agentic AI | It is mainly used for Transactional Interactions |
Complexity | It supports various conversations and Agentic AI responses | It supports single input query |
Response Type | Automatically generates responses | Initiates a system process based on the input, but without an instant response |
Prompt
A prompt is an input provided to a language model to guide its response. It provides structured instructions and relevant variables to ensure consistent and accurate outputs. Essentially, a prompt directs the model to perform a specific task or answer a query.
Prompt components are the building blocks for creating structured, reusable prompts. They allow dynamic customization for different scenarios, ensuring the model responds accurately to changing variables or user inputs.
Example
Let's look at a real-world example where prompts are used in Support Team:
Template
"Hi, {{customer_name}}, your i_ticket {{ticket_number}} is under processing. We will get back to you by {{response_time}}."
Filled Template
"Hi, Sam, your i_ticket #61587 is is under processing. We will get back to you by 4:00 PM today."
A sample screenshot is shown below:
Figure: Prompt - Agent Instruction
Template: The Template field defines the pre-defined structure and format of the prompt, providing a blueprint for generating consistent outputs. It combines static text and dynamic variables.
user_details: The User_details Field provides the dynamic user-specific information that personalizes the prompt, allowing the AI model to produce contextually relevant and customized responses.
Example:
By combining both, you can create prompts that are not only structured and repeatable also.
Example
Template Field:
"Hello {{user_name}}, your booking for {{destination}} is confirmed for {{travel_date}}."User_details Field:
{"user_name": "Alice", "destination": "Paris", "travel_date": "March 15, 2025"}
Orchestration Agent
An Orchestration Agent is an AI-driven system that automates tasks by following predefined instructions and integrating AI models, APIs, and tools. It requires key configurations like model provider details, API keys, and task instructions to operate efficiently. A sample screenshot is shown below:
Figure: Orchestration Agent
The following table describes the fields and its description.
Field | Description |
|---|---|
Model Provider | Select the AI model provider such as Azure OpenAI, OpenAI, that the agent will use for generating responses. |
API Key | Specifies a unique API Key which is a secure credential used to authenticate and authorize access to an AI model provider’s API. |
Agent Instructions | Specifies how the Orchestration Agent should behave, what tasks it must perform, and how it should interact with inputs and tools. These instructions act as a guiding framework for the agent, ensuring it follows predefined rules while making autonomous decisions. |
Tools | These instructions serve as a blueprint for the agent, enabling it to operate autonomously while adhering to predefined rules. |
Input | The data or user prompt that the agent processes to generate an output. |
API Version | Select the version of the API the Orchestration Agent should use when interacting with the AI model provider. |
Deployment Name | It is a unique identifier assigned to a model that allows the Orchestration Agent to communicate with the correct instance for processing requests. |
Azure Endpoint | Specifies the URL that enables seamless integration with Azure AI services for processing data requests. |
Tools
Tools are specialized functions or external resources that an agent can call to perform tasks. These tools are typically linked to the agent through a Tools port and are registered during agent initialization, allowing the agent to recognize and utilize them as needed. A sample screenshot is shown below:
Figure: Tools - Display Chart
Each tool is wrapped as a Tool block with a standardized interface, making it easier for the agent to interpret and execute functions. The Tool’s description provides details on its capabilities, enabling the agent to make informed decisions.
How Agents Use These Tools:
The agent registers available tools during initialization.
When a user request comes in, the agent determines which tool is needed.
The agent calls the selected tool, executes the function, and returns results.
Chat Output
Chat Output component is fundamental for conveying the processed results to the user, after processing the input from the End User. It helps maintain the conversational flow by showing responses that are contextually relevant to the user's inputs.
Example
Chat Input - “How can I get to know my leave balance?
Chat Output - “To know your leave balance, perform the following steps:
1. Log in to HRMS application.
2. Enter your employee ID.
3. Click Leave Balance.”
A sample screenshot is shown below:
Figure: Chat Output
Was this article helpful?