Core Concepts
Last updated: Jan 2026
Overview
Before diving into building complex automations, it's important to understand the core concepts that make ORCFLO work. These foundational ideas will help you design better workflows and troubleshoot issues effectively.
Workflows
The container
Nodes
The building blocks
Connections
The data flow
Tools
The capabilities
What is ORCFLO?
ORCFLO is a no-code AI workflow automation platform. It allows you to build sophisticated AI-powered automations visually, without writing code. Simply connect nodes on an intuitive visual canvas and data flows automatically between them.
Key Features
- Visual Builder: Drag-and-drop interface for building workflows
- Multi-Model AI: Use GPT, Claude, Gemini, and more in one workflow
- Automatic Data Flow: Data passes between connected nodes seamlessly
- Real-time Execution: Live progress updates and debugging
Workflows
A workflow is the container for your automation. It's a visual representation of a process that takes inputs, performs operations, and produces outputs. Think of it as a recipe that ORCFLO follows to accomplish a task.
Workflow Characteristics
- Executable: Run manually or on schedules
- Composable: Built from reusable nodes and patterns
- Configurable: Accept different inputs for each run
- DAG or Sequential: Choose parallel or step-by-step execution
Nodes
Nodes are the building blocks of your workflow. Each node represents a single operation. ORCFLO provides 5 different node types organized into 3 categories.
Data Nodes
| Node | Description |
|---|---|
| Input | Define workflow entry points |
| Output | Collect and format results in markdown |
AI Nodes
| Node | Description |
|---|---|
| LLM Step | Connect to AI models (Claude, GPT, Gemini) with task instructions |
| Decision Point | AI-powered decision-making that routes to different branches |
Utility Nodes
| Node | Description |
|---|---|
| Note Card | Add annotations and documentation directly on the canvas |
Node Configuration
Click on any node to open its configuration panel on the right side of the screen. This is where you set parameters like task instructions, models, and output formats.Connections
Connections (also called edges) link nodes together and define the flow of execution. When a node completes, its output is passed along the connection to the next node.
Connections carry data between nodes. The output of one node becomes available as input to the connected nodes. This creates a pipeline where data is transformed at each step.
- Connections are created by dragging from one node's output handle to another's input
- A node can have multiple outgoing connections (for branching)
- A node can receive inputs from multiple sources
- The execution order follows the connection paths
Tools
Tools extend your AI nodes with real-world capabilities. When you enable tools on an LLM Step, the AI can autonomously decide when to use them to accomplish its task. This transforms simple prompts into powerful agentic workflows.
Agentic AI
Tools enable agentic behavior. The AI doesn't just generate text, it takes actions. It can search the web, send emails, make API calls, and more, all based on the context of your prompt.Available Tools
| Tool | Description |
|---|---|
| Web Search | Search the internet for real-time information and data |
| HTTP Requests | Make API calls to external services and endpoints |
| Send Email | Send emails programmatically as part of your workflow |
| Document Processing | Read, parse, and extract data from files |
Triggers
While tools extend what AI can do during execution, triggers determine when your workflow runs. Triggers can start workflows automatically.
| Trigger | Description |
|---|---|
| Scheduled | Run on a schedule (hourly, daily, weekly, or custom cron) |
| Webhook (coming soon) | Trigger via HTTP request from external services |
Adding Tools
To add tools to an LLM Step, drag and drop specific tools from the component library onto the LLM Step, or add tools directly in the LLM Step configuration panel.Execution
Execution is when your workflow runs. You can trigger execution manually (Execute button) or on a schedule. ORCFLO provides real-time visual feedback during execution.
Start
Workflow begins execution
Process
Nodes execute based on execution mode (DAG or Sequential)
Visual Feedback
Nodes turn green (success) or red (failure) in real-time
Complete
Each step has its own individual output with execution metrics
During execution, you can monitor progress in real-time. The execution inspector shows inputs, outputs, and metrics (tokens, time, cost) for each node.
Putting It All Together
Let's see how these concepts work together in a real workflow example: a customer support email classifier.
Customer Email Classifier
- Input: Receives the customer email text via input fields
- LLM Step: Claude analyzes sentiment and extracts key topics
- Decision Point: AI-powered routing based on urgency level
- Output: Returns category, priority, and suggested response in markdown
Best Practices
- Start simple: Begin with linear workflows before adding branches and loops
- Name nodes clearly: Use descriptive names so variable references are readable
- Test incrementally: Run your workflow after each change to catch issues early
- Use the inspector: Inspect node inputs/outputs and metrics to understand data flow
Key Takeaways
- ORCFLO is a no-code AI workflow platform with a visual drag-and-drop builder
- Workflows support DAG (parallel) or Sequential execution modes
- 8 node types across 4 categories: Data (Input, Output), AI (LLM Step, Decision Point), and Utility (Note Card)
- Tools enable agentic AI behavior: web search, HTTP requests, email, and document processing
- Triggers (scheduled) automate when workflows run