Setting Up and Using AI Agent Studio in Captivation Hub

Agent Studio is Captivation Hub's no-code/low-code platform for designing, managing, and deploying AI-driven agents. Between the drag-and-drop builder, the tool nodes, and flexible deployment options, you can build intelligent agents without writing code — whether you're putting together a conversation flow, hooking into APIs, or attaching a knowledge base.

Note: Agent Studio is available to Agency Admins/Users and Sub-account Admins/Users.

What is Agent Studio?

Agent Studio is a visual platform inside Captivation Hub for creating AI-powered agents. It pairs a drag-and-drop workflow builder with versatile tool nodes and built-in integrations so you can design agents that respond intelligently, automate processes, and connect with external data sources.

The result is a way to build agents that handle customer support, streamline operations, and slot into the systems you already use.

Key Benefits

• Drag-and-drop builder: Create workflows quickly with an intuitive visual canvas.
• Versatile tool nodes: Add intelligence, knowledge, and integrations through dedicated node types.
• Logic and flow control: Manage branching, conditions, and variable handling for smarter automation.
• Version control: Develop, test, and deploy agents through structured lifecycle stages.
• Built-in testing: Validate agent performance before pushing changes to production.
• Integration flexibility: Connect to APIs, knowledge bases, and the web for context-aware responses.
• Export/import: Back up, share, or migrate agents between environments.
• API access: Run agents programmatically inside your existing workflows.

The Drag-and-Drop Agent Builder

The visual builder is where you create and manage your AI agent workflows. Connect nodes, set flow conditions, and design complex processes without leaving the canvas.

From the builder you can:

• Connect tool nodes for logic, data, and intelligence
• Configure workflows visually without writing code
• Manage runtime and global variables for dynamic responses

Tool Nodes

Tool nodes are the core building blocks of Agent Studio. Each node type adds a specific capability so you can shape an agent around exactly what you need.

LLM Node

Uses advanced large language models to generate intelligent, context-aware responses. You control prompts, parameters, and how the output is handled.

MCP Tool Node

Integrates external tools through the Model Context Protocol. Useful when you need to access outside data sources or trigger actions in another system.

API Tool Node

Connects directly to third-party services by sending and receiving data via APIs. This lets the agent interact with outside systems on the fly.

Knowledge Base Tool Node

Attach a knowledge base so the agent can answer based on uploaded documents or stored information. Great for keeping responses accurate and consistent.

Web Search Tool Node

Lets the agent pull current information from the internet. Useful when answers need to be fresh or rely on publicly available data.

Logic and Flow Control

Logic and flow control let your agent handle complex scenarios and make decisions based on conditions:

• Branching via edge conditions: Define different paths depending on variable values or input.
• Variable management: Create global, runtime, or input variables for context-aware, flexible responses.

Version Control

Agent Studio supports structured lifecycle management with Staging and Production modes. You can test changes safely before they affect live conversations.

• Staging: Your work-in-progress version. Saving updates the staging copy without touching live traffic.
• Production: The live version that runs for real contacts. A production version is created only when you promote a staging version.

For details on promoting, restoring, and working with multiple production versions, refer to the related Agent Studio version management article.

Export and Import Agents

Export and import lets you back up, share, or move agents between environments. It makes collaboration and migration much easier — no rebuilding from scratch.

Test Tab

The built-in Test tab lets you simulate and validate workflows before pushing them live. Use it to catch errors, confirm logic paths, and refine performance.

API Access

Agent Studio exposes secure APIs so you can execute your agents programmatically. This is what lets you embed agents into your own systems or external workflows.

How to Set Up Agent Studio

1. Navigate to AI Agents → Agent Studio from your Captivation Hub dashboard.
2. Click Create Agent to open the drag-and-drop builder.
3. Add tool nodes — LLM, MCP, API, Knowledge Base, or Web Search — to your workflow.
4. Connect nodes with edges to define logic paths.
5. Configure variables to pass data between nodes.
6. Use the Test tab to validate the agent's flow.
7. Save your agent to create or update the staging version. When you're ready, promote staging to production.
8. (Optional) Export or share your agent if needed.
9. Use API Access to run agents programmatically.

Frequently Asked Questions

How is Agent Studio different from Conversation AI?
Agent Studio gives you a visual builder with multiple node types, which makes it better suited for complex workflows and integrations than Conversation AI alone.

Can I use my own documents in Agent Studio?
Yes — attach a knowledge base with your uploaded documents and the agent will use them to deliver precise, source-backed answers.

What is MCP, and why would I use it?
MCP (Model Context Protocol) lets you integrate external tools, expanding what your agent can do beyond the standard nodes.

Is version control mandatory when creating agents?
It's not required, but it's recommended. Staging and production versions let you test safely before deploying changes to real conversations.

Can I migrate agents between accounts?
Yes — the export/import functionality lets you share agents across accounts or environments without rebuilding from scratch.