Terminology and key concepts
Amazon Quick Flows empowers business users to transform their everyday tasks into workflows for individual and team productivity. Understanding the core terminology and concepts helps you effectively create, run, share, and maintain Flows within your organization. Flows offer agentic runtime support enabling end users to chat with their workflow and request updates like summarizing an output or skipping a step.
User inputs
User inputs represent the entry points where users provide information to initiate a flow. These steps capture the necessary context from a user that is needed to execute the flow. Input steps can include text fields for user queries or file upload capabilities for document processing. The design of input steps directly impacts user experience and determines how effectively users can communicate their needs to the flow.
When configuring user inputs, consider the type of information your flow needs to process and choose appropriate input methods that align with user expectations. Well-designed user inputs provide clear guidance about what information is required through placeholder text and some default values, if applicable.
Note
Eligible steps can be referenced in subsequent steps through the @ reference notation. Type @ in prompt to see menu of eligible steps within that flow.
AI-generated responses
AI-generated responses are outputs that define how Amazon Quick Flows presents results and information back to users. These steps encompass various formats including text responses, AI-generated images using Bedrock models, or structured data presentations as Amazon Quick Sight visuals. Adding a specific step for AI-generated response denotes what data will be used to produce output for that step, which will be available later in the flow for further processing. These outputs can be generated from connected knowledge bases, Amazon Quick Sight dashboards and topics, spaces defined in Quick Suite, or results from the web.
Image generation capabilities allow Flows to create visual content dynamically, creating customized, consistent images for ads, marketing assets, stock photos, social media, and e-commerce content. These AI-powered image outputs can be tailored to specific requirements and integrated seamlessly with other output formats within the same Flow.
Action steps
Action steps enable Amazon Quick Flows to interact with external systems and perform automated tasks beyond content generation. These steps connect your flows to third-party applications, AWS services, and internal systems through a comprehensive library of pre-built connectors. Action steps can perform tasks such as sending notifications, updating databases, creating calendar events, or any custom action defined with MCP or an OpenAPI specification (learn more about Quick actions in Action steps in flows).
The extensive action connector library supports integration with popular business applications, development tools, and AWS services. When implementing action steps, consider authentication requirements, data security implications, and the specific permissions needed for your intended integrations.
Reasoning groups
Reasoning groups organize related processing logic within Amazon Quick Flows, allowing creators to control how the flow executes, using natural language instructions. These groups help manage the flow of information and decision-making processes by grouping related steps that work together to accomplish specific objectives. Reasoning groups improve workflow organization, making Flows easier to understand, maintain, and troubleshoot.
Within reasoning groups, you can establish dependencies between steps, control execution order, and manage data flow between different components of your workflow. This organizational structure becomes particularly valuable when building flows that require multiple processing stages or conditional logic. Instructions within reasoning groups provide specific guidance to define how the flow should interpret inputs, process information, and generate appropriate responses.
Editor and Run mode
Amazon Quick Flows provides two distinct modes for working with flows: Editor mode for building and configuring flows, and Run mode for executing and testing them. You can switch between these modes using the mode toggle in the interface.
In Editor mode, you design your flow by adding and configuring steps, defining logic, and setting up the workflow structure. This is where you build the automation before making it available to users.
Run mode allows you to execute and test your flow. Within Run mode, you can interact with your flow using two different interaction patterns: chat mode for conversational, iterative interactions where users can ask follow-up questions and refine their requests, or structured mode for guided, step-by-step workflows that follow predetermined paths and ensure consistent data collection.
Getting started with terminology
Understanding these core concepts provides the foundation for effectively using Amazon Quick Flows. As you begin building flows, refer back to these definitions to ensure you're leveraging the appropriate components for your specific use cases and organizational requirements.
