Skip to content

Automation Mode

Automation Mode is a powerful feature in Wabee AI Studio that provides deterministic control over agent execution through structured JSON configurations. Unlike traditional autonomous agents, Automation Mode gives you precise control over task sequencing, dependencies, and argument passing - perfect for building reliable agentic automations.

Overview

In Automation Mode, agents execute predefined workflows specified through a JSON configuration. This approach is ideal when you need:

  • Predictable execution paths - Know exactly what your agent will do
  • Complex task orchestration - Define dependencies between tasks
  • Controlled tool usage - Specify which tools can be used for each task
  • Variable passing - Share outputs between tasks using dynamic variables
  • Pre-configured arguments - Set known values upfront to avoid LLM inference

Key Concepts

Workflow Plans

A workflow plan defines the complete automation sequence, including: - Tasks - Individual units of work with specific objectives - Dependencies - Control task execution order - Variables - Pass data between tasks - Constraints - Optional execution rules

Task Types

Automation Mode supports two main task types:

  1. Standard Tasks - Execute a single tool or action
  2. React Loop Tasks - Run multiple iterations using available tools until completion

Dynamic Variables

Variables enable data flow between tasks. Any task output can be stored in a variable and referenced by subsequent tasks using the $$variable_name$$ syntax.

Known Arguments

Pre-specify tool arguments to ensure consistent execution without relying on LLM inference. This is crucial for automation reliability.

When to Use Automation Mode

Choose Automation Mode when you need:

Repeatable processes - Same execution path every time
Compliance requirements - Auditable, deterministic workflows
Integration workflows - Predictable API calls and data transformations
Quality assurance - Consistent output formats and validation
Cost optimization - Reduced LLM calls through pre-configured arguments

Automation Mode may not be suitable for:

Exploratory tasks - When the path isn't known upfront
Creative work - Tasks requiring flexibility and adaptation
Unpredictable scenarios - Situations with many unknown variables

Getting Started

To use Automation Mode:

  1. Navigate to the agent creation or update page in Wabee AI Studio
  2. Select "Automation Mode" as the agent type
  3. Paste your JSON configuration in the provided editor
  4. Validate and save your automation

Next Steps