TypeScriptADK-TS

Workflow Agents

Orchestrate multiple agents with predictable execution patterns

Workflow agents provide deterministic orchestration patterns for coordinating multiple agents. Unlike LLM agents that make dynamic decisions, workflow agents follow predictable execution paths.

Workflow Types

When to Use Workflow Agents

Workflow agents are ideal for:

  • Predictable pipelines - When you need guaranteed execution order
  • Data processing workflows - Multi-step data transformation and analysis
  • Quality assurance - Review and validation workflows
  • Performance optimization - Parallel processing of independent tasks
  • Iterative improvement - Repeated refinement until criteria are met

Key Benefits

Deterministic Execution

Unlike LLM agents, workflow agents provide predictable, repeatable execution patterns.

Composition

Combine different workflow types to create complex orchestration patterns.

Error Handling

Built-in error handling and recovery mechanisms for robust workflows.

Performance Control

Optimize execution with parallel processing and conditional loops.

Workflow Patterns

Common Patterns

  • Research Pipeline: Sequential research → analysis → summary
  • Content Review: Parallel fact-checking and grammar review
  • Iterative Refinement: Loop until quality criteria are met
  • Distributed Processing: Parallel execution across data partitions

Choosing the Right Workflow

  • Sequential: When order matters and each step depends on the previous
  • Parallel: When tasks are independent and can run simultaneously
  • Loop: When you need iterative improvement or retry logic

Advanced Features

Nested Workflows

Workflow agents can contain other workflow agents for complex patterns.

Conditional Execution

Control flow based on agent outputs and session state.

Result Aggregation

Combine results from multiple agents into coherent outputs.

State Management

Pass data between agents using session state and context.

Getting Started

Start with the workflow type that matches your use case:

  1. Sequential Agents - For step-by-step processes
  2. Parallel Agents - For independent tasks
  3. Loop Agents - For iterative processes