Workflows allow you to chain multiple AI agents together to automate complex, multi-step processes. Instead of manually coordinating between agents, workflows handle the orchestration automatically.

What are Workflows?

A workflow is an automated sequence of AI agent tasks that work together to accomplish a goal. Each task in a workflow is handled by a specific agent, and the output from one task can flow into the next.

Example use cases:

Content creation pipeline: Research → Write → Edit → Publish
Lead processing: Qualify → Enrich → Score → Route
Support ticket handling: Categorize → Research → Draft Response → Review

Workflow Types

AffinityBots supports two types of workflows:

Sequential Workflows

Tasks execute in a linear chain, one after another. The output from each task flows to the next in sequence.

Trigger → Task 1 → Task 2 → Task 3 → Output

Best for:

Predictable, step-by-step processes
Content pipelines with clear stages
Data transformation workflows

Example: A blog post workflow where:

Research Agent gathers information on a topic
Writer Agent creates a draft
Editor Agent refines and polishes the content

Orchestrator Workflows

A manager agent dynamically decides which worker agents to use and in what order. The manager analyzes the goal, delegates tasks, and synthesizes results.

Trigger → Manager Agent ⟷ Worker Agents → Output

Best for:

Complex problems requiring adaptive decision-making
Tasks where the best approach isn't predetermined
Situations requiring judgment about which agents to involve

Example: A customer inquiry workflow where:

Manager Agent analyzes the question type
Delegates to appropriate specialist (Technical Support, Billing, Sales)
Synthesizes the response before sending

Key Concepts

Triggers

Triggers define how a workflow starts. Options include:

Manual: Start via UI or API
Webhook: Start when external system sends data
Schedule: Run on a cron schedule (hourly, daily, etc.)
Form: Collect user input before starting
Integration: Start from connected apps (coming soon)

Tasks

Tasks are the work units within a workflow. Each task:

Is assigned to a specific AI agent
Has configurable context settings
Can include custom instructions
Produces output for the next step

Context Control

Control how much information flows between tasks:

Full Context: Agent sees entire conversation history
Isolated: Agent starts fresh with just the input
Custom Instructions: Additional guidance per task

Thread Management

Choose how conversations are managed:

Shared Thread: All tasks share context
Separate Threads: Each task has isolated context

When to Use Each Type

Scenario	Recommended Type
Fixed, predictable process	Sequential
Dynamic, adaptive process	Orchestrator
Simple automation	Sequential
Complex decision-making	Orchestrator
Content pipeline	Sequential
Customer service routing	Orchestrator

Getting Started

Ready to build your first workflow? Head to the Creating a Workflow guide to get started.

Related Guides

Creating a Workflow - Step-by-step workflow creation
Workflow Triggers - Configure how workflows start
Adding Guardrails - Validate and filter agent outputs

Introduction to Workflows