AI Agent Ports — Agentic AI

AI Agent Workflow 1 — BizFirstAI Studio showing the AI Agent node with all port types and the LLM and Memory satellites on the Manual Trigger

AI Agent Workflow 1 — AI Agent node with all six port types visible · Click image to view fullscreen

Port Diagram

AI Agent node — port connections overview

flowchart LR IN([Input]) --> AG[AI Agent\nOctopus] AG --> OUT([Output]) BA([Base Agent]) -. It Extends .-> AG SA([Sub Agent]) -. Delegation .-> AG CO([Component]) -. Composite .-> AG AG -. My Components .-> MC([Components])

Port Types

Ports fall into two categories: Runtime ports carry data during execution, and Static ports define structural relationships that are resolved at design time before the workflow runs.

Runtime

Input

Receives data flowing into the agent from upstream nodes in the workflow. All execution data arrives through this port.

Runtime

Output

Emits the agent's result downstream. All nodes that depend on this agent's reasoning receive data through this port.

Static

It Extends

Connects this agent to a parent base agent. The current agent inherits the parent's configuration — LLM settings, tools, memory — at design-time resolution.

Static

Sub Agent

Attaches a subordinate agent for task delegation. The main agent orchestrates its sub-agents, assigning subtasks and aggregating results.

Static

Component

Used to attach composite child services to this agent. For example, combining a document server and email server into a single composite agent capability.

Static

My Components

Exposes this agent's own composite component slots — the services it owns and makes available to other nodes that connect to it as a component.

Pre-Connected Satellites on Manual Trigger

In the screenshot, the Manual Trigger node already has two satellite nodes wired — these are passed through to the AI Agent as its backing services:

Satellite	Purpose
LLM	The language model provider (e.g. OpenAI, Azure OpenAI) that backs the agent's reasoning. Defines model, temperature, and token limits.
Memory	The memory store the agent reads from and writes to during execution. Enables multi-turn context, session memory, or persistent knowledge.

Agent Internal Capability Tabs

The AI Agent node surfaces three internal tabs at the bottom of the node panel, each configuring a different aspect of the agent's capabilities:

Tab	Description
Tools	Tool-calling integrations the agent can invoke — APIs, functions, database queries, external services.
Base Agent	The parent agent this agent inherits from via It Extends. Shows the resolved inheritance chain.
Sub Agents	Child agents orchestrated under this agent. Each sub-agent handles a delegated subtask.

Port Relationship Reference

Port	Category	Resolved When	Purpose
Input	Runtime	Execution	Receives upstream workflow data
Output	Runtime	Execution	Emits results downstream
Component	Static — Composite	Design time	Attaches child service components
It Extends	Static — Inheritance	Design time	Inherits base agent configuration
Sub Agent	Static — Delegation	Design time	Nests subordinate agents
My Components	Static — Composite	Design time	Exposes owned component slots

Design-Time Resolution All static ports (Component, It Extends, Sub Agent, My Components) are resolved before runtime. BizFirstAI collapses the full inheritance and composition tree into a single unified configuration for each agent, so runtime execution sees a flat, consolidated agent — not a graph of relationships.

01 — Fetch & Insert Workflow 03 — Deep AI Agent