Automation Concepts
Before building your first automation, it’s important to understand the core concepts that power Agent Studio’s workflow engine.Nodes (Components)
Nodes are the building blocks of automations. Each node performs a specific function and can be connected to other nodes to create a workflow.Node Categories
| Category | Purpose | Examples |
|---|---|---|
| Triggers | Start the automation when an event occurs | Account Segment, Task Created, Scheduler |
| Processing | Transform, filter, or combine data | Create Data, Update Data, Filter Data |
| Logic | Route data based on conditions | If-Else, Condition |
| Actions | Perform operations like sending notifications | Send Email, Slack Message, Create Task |
| AI | Use LLMs for intelligent processing | Agent, LLM, Structured Output |
Data Flow
Data flows through your automation from node to node via connections. Each node receives data from its inputs, processes it, and passes results to connected outputs.The Data Object
All data in Agent Studio flows as Data objects - structured containers with key-value pairs:Accessing Data with @ Notation
When configuring nodes, you can reference values from incoming data using the@ notation:
| Syntax | Description | Example |
|---|---|---|
@key | Access a top-level key | @account_name → “Acme Corp” |
@key.nested | Access nested values | @account.owner.email |
@ | Access the entire data object | Full data dictionary |
Connections
Connections define how data flows between nodes. To connect two nodes:- Click on a node’s output handle (right side)
- Drag to another node’s input handle (left side)
- Release to create the connection
Connection Rules
- Every automation must start with exactly one Trigger node
- Nodes can have multiple inputs and outputs
- Logic nodes (If-Else, Condition) have separate True and False outputs
- Actions always execute when reached, even if not connected to downstream nodes
Flow Context
When a trigger fires, it establishes a Flow Context that provides entity information throughout the automation:| Context Field | Description |
|---|---|
account_id | The account associated with the trigger |
account | Full account data object |
task_id | The task (if applicable) |
task | Full task data object |
user_id | The user/contact (if applicable) |
activity_id | The activity (if applicable) |
Execution Model
Node Execution
When a flow runs:- The trigger node executes first, loading entity data
- Connected nodes execute in order, following the data flow
- Each node processes its inputs and produces outputs
- Important: Action nodes always execute their side effects (emails, API calls, etc.) even if their outputs aren’t connected to anything
Conditional Routing
Logic nodes like If-Else and Condition split the flow into multiple paths:Error Handling
If a node fails during execution:- The error is logged with details
- Downstream nodes on that path do not execute
- Other parallel paths continue normally
- The flow status reflects the failure
Best Practices
Start simple, then add complexity
Start simple, then add complexity
Begin with a basic trigger → action flow. Once working, add processing and logic nodes incrementally.
Test nodes individually
Test nodes individually
Use the play button on each node to test it in isolation before running the full flow.
Use meaningful names
Use meaningful names
Rename nodes to describe their purpose (e.g., “Check Health Score” instead of “Condition V2”).
Document with comments
Document with comments
Add description fields to complex nodes explaining their purpose and configuration.
Handle edge cases
Handle edge cases
Consider what happens with missing data, empty lists, or unexpected values. Use conditions to handle these cases gracefully.
Next Steps
Build Your First Automation
Follow a step-by-step tutorial
Explore Triggers
Learn about available trigger types