A Beginner Walkthrough of Agent Studio

Agent Studio is the visual environment inside Captivation Hub where you build intelligent AI agents. You design how an agent gets triggered, how it processes information, how it makes decisions, and what actions it performs — all on a flexible, node-based canvas.

This article is the starting point if you're exploring Agent Studio for the first time. It walks through the dashboard, the canvas, and the building blocks you'll use before moving on to more advanced builds.

Quick note: Agent Studio and AI Studio are two different tools inside Captivation Hub. AI Studio is a conversational, AI-powered builder for creating websites, pages, and other web assets from prompts. Agent Studio — what this article covers — is a visual builder for AI agents that respond to events, interact with users, and automate tasks.

Agent Studio Dashboard

The dashboard is the home base for everything you build. Open AI Agents > Agent Studio and you'll see every agent in your account.

From the dashboard you can:

• View existing agents.
• Check whether each one is in Draft or Published status.
• See when an agent was created or last updated.
• Organize agents into folders.
• Create a new agent.
• Browse the available templates.

Creating a New Agent

Click Create Agent from the dashboard. You'll have two options:

• Start from scratch — opens a blank canvas.
• Use a template — installs a pre-built structure from the Template Library that you can customize.

The Agent Template Library

The Template Library holds ready-made agent blueprints for common use cases. Instead of starting from a blank canvas, you can browse the categorized templates, preview how each one is structured, install one into your account, and modify any part of the flow afterward.

Templates typically come pre-wired with triggers, AI nodes, routing logic, and actions. For first-time users, they're often the fastest way to understand how agents are connected together.

The Builder Canvas

Open or create an agent and you land in the visual builder canvas. This is where you drag and connect nodes, define execution logic, and configure each step of the agent.

Agents flow left to right — they begin with a trigger and end with an action or completion point. The connections between nodes define how the agent behaves in real conversations.

The Start Trigger

Every agent begins with a Start Trigger. The trigger defines when the agent should run. Common triggers include:

• Chat Message
• Form Submission
• Lead Tag Added

Without a trigger, the agent will not execute. The trigger is what tells the agent which event should kick it off.

Node Categories

Everything on the canvas is built using nodes. A node is a building block that performs one specific function — starting the agent, generating a response, collecting information, or connecting to another system. You connect nodes with edges to define logic paths.

The main node types you'll work with:

• AI Agent — the brain of the flow. It reads the incoming input, generates responses, and uses tools like a Knowledge Base or Web Search.
• Sequential — groups multiple steps that should run together as a single sequence.
• End Node — marks where the agent's execution stops. Once the flow reaches an End Node, the agent's run is complete.
• Router — sends the conversation down different paths based on context. Think of it as a decision point: if the user asks about support, go one way; if they ask about sales, go another.
• Search Knowledge Base — looks up information stored in your connected Knowledge Base, so the agent can pull answers directly from your documentation.
• Search Web — gathers information from online sources when responding to questions that need publicly available or current data.
• MCP Server — connects the agent to supported external tool servers. Used for advanced integrations where the agent needs to call into outside systems.
• API Call — sends data to or pulls data from another application. Useful for pushing lead data into another system or retrieving customer details on demand.
• Audio Generation — produces spoken audio responses based on instructions.
• Email Address — collects a structured email from the user during the interaction.
• Phone Number — collects a phone number in a standardized format.
• Single Choice — presents predefined options for the user to pick from.
• Text Input — collects open-ended responses for richer information gathering.

Variables and the Global Prompt

At the top of the builder you'll find configuration tools that govern how the agent behaves across the entire flow.

Variables

Variables are reusable placeholders. Define a value once and reference it anywhere in the agent — for example, store a business name once and let every response pull from the same variable instead of hard-coding it in multiple places.

Global Prompt

The Global Prompt sets overarching instructions that guide the agent in every interaction — tone, personality, rules to follow, and any guardrails you want consistently applied.

Testing and Publishing

Before activating an agent, you can:

• Use the Test feature to simulate execution and see how the agent reacts.
• Save changes as a draft.
• Publish the agent when you're ready.

An agent has to be published before it will respond to real-world triggers.

Getting Started — A Quick Path

To create your first agent:

1. Open AI Agents > Agent Studio.
2. Click Create Agent.
3. Choose to start from scratch or install a template.
4. Add a Start Trigger.
5. Drag an AI Agent node onto the canvas and connect it.
6. Use Test to simulate behavior.
7. Save and Publish.

Frequently Asked Questions

Q: Do agents run automatically?

Only when their configured trigger conditions are met. If the trigger doesn't fire, the agent stays idle.

Q: How is Agent Studio different from regular workflows?

Workflows follow predefined automation rules. Agents can interpret input and make dynamic decisions on the fly using AI.

Q: Do I have to publish an agent for it to work?

Yes — an agent has to be published before it can respond to real events.

Q: Can agents integrate with external systems?

Yes. The API Call and MCP Server nodes both let the agent reach into outside tools.

Q: Can agents generate things beyond text?

Yes. Agent Studio supports image, audio, and video generation.