AWSAWS Certified Generative AI Developer - Professional2026-102.5 Implement application integration patterns and development tools
Choose workflows by their operating owner
Completed by the knowledge check
Learning objective
Choose Bedrock Prompt Flows for developer-managed orchestration, Amazon Quick for business research and automation, or Amazon Q Business Apps for business-user applications
Core idea
Choose the product from who builds, releases, and runs the workflow. A developer-managed application needs an invocation contract and a controlled deployment target. A business research workflow needs an environment where users can select sources, produce reports, and connect subsequent actions. A reusable business-user app needs an appropriate application environment and access model.
AWS documentation calls the first product Amazon Bedrock Flows, rather than Bedrock Prompt Flows. It calls purpose-built apps within Q Business Amazon Q Apps, rather than Amazon Q Business Apps. Preserve those distinctions when translating the objective into a service choice: similarly named flows and apps do not share one runtime.
There is also a service-availability constraint. Amazon Q Business is no longer open to new customers. AWS continues bug fixes and security updates for existing customers but does not consider new feature requests, and it recommends Amazon Quick for new generative or agentic solutions. Learn Q Apps as an existing-customer integration option, not as an unrestricted choice for a new deployment.
How it works
Put developer-managed inference under Bedrock Flows
Amazon Bedrock Flows connects prompts, supported foundation models, knowledge bases, and services such as AWS Lambda. The developer defines nodes and connections, gives the flow appropriate permissions, and integrates its runtime into an application. A visual builder changes how you author that definition; it does not transfer production ownership to the end user.
The deployment mechanism separates an editable working draft from an immutable version. After testing a draft, publish a version and create an alias that points to it. Application callers use InvokeFlow against that alias. Updating the alias’s routing configuration changes the deployed target without requiring each caller to discover the version number.
This is a useful boundary when a product team owns a repeatable inference pipeline and wants to review what reaches production. A draft edit should not silently become the behavior of an application using a released version. Detailed node typing and conditional routing belong to 1.6.8–1.6.9; the product choice here turns on deployment ownership and application invocation.
Choose the flow’s dependent resources deliberately. Connecting a Lambda node does not make an external write safe by itself. The application still needs authorized inputs, bounded actions, and evidence that the downstream operation succeeded. A published definition preserves the flow snapshot; release evidence should also identify the dependencies it invokes.
Put business research and automation in Amazon Quick
Amazon Quick combines research, task automation, analytics, and app building in a business-facing environment. Quick Research conducts research across the web and connected data and delivers a cited report. Quick Flows lets users create, customize, and share sequences that gather inputs, generate responses, apply logic, and take actions in connected applications.
The concrete composition mechanism is a flow’s step type. A Research step invokes Quick Research; an Application actions step reads or writes to connected applications; Text and Files steps collect user input. A business workflow can therefore connect an account-research result to a later action instead of requiring a developer to turn every report into a custom service endpoint.
Define which sources the research should use and which downstream actions the business process permits. Keep review of the report separate from authorization for a write.
Quick Automate is another capability within Quick for business process automation with agents that make contextual decisions and execute actions across applications. Do not use the presence of “automation” in a requirement to assume that Quick Flows, Quick Automate, and Bedrock Flows are interchangeable. Identify whether you need a user-created sequence, a business process automation, or a developer-invoked inference graph.
Keep Q Apps within their supported environment
Amazon Q Apps lets users turn useful Q Business conversations into lightweight, reusable, shareable applications using enterprise data. The documented card model includes input cards, text output cards, plugin cards, and file upload cards. A marketing content app can collect product information and generate several approved-format drafts; an onboarding app can combine inputs, a Jira plugin action, and a welcome-message output.
Q Apps requires Amazon Q Business Pro users; Lite users cannot create, run, or view these apps. The Q Apps capability lives within the broader Q Business application environment and is accessed through its web experience. Its programmatic management APIs do not turn it into an independent replacement for a Bedrock application runtime.
For an existing Q Business customer, start by checking the actual user tier and application environment. Preserve access and action controls while assessing migration. For a new customer, the closure of Q Business is the deciding constraint. Assess Apps in Amazon Quick, which builds interactive web applications from natural-language descriptions. This capability does not establish unchanged migration of existing Q Apps.
AWS’s Q Business migration guidance describes bringing an existing Q Business index into Quick through Bring Your Own Index (BYOI), with account, Region, and identity requirements. It also directs customers using anonymous access or API integration into custom applications to AWS Support for migration approaches. Do not promise that an existing app or API integration moves unchanged merely because Quick offers similar capabilities.
Key distinctions and trade-offs
The products differ in operating boundary:
| Requirement | Relevant mechanism | Ownership decision |
|---|---|---|
| Released inference graph invoked by application code | Bedrock flow version, alias, InvokeFlow | Developer owns the application integration |
| Research feeding repeatable business actions | Quick Research and Quick Flows steps | Business workflow runs in Quick |
| Reusable card-based app in an existing Q Business environment | Amazon Q Apps and Pro subscription | Existing environment and user eligibility govern access |
A visual editor is not the deciding feature: both developer and business products can offer visual authoring. Likewise, a cited research report is an evidence artifact, not automatic authorization to change a customer record.
Managed composition reduces the code you write, but it places behavior inside the selected product’s configuration and permissions. Confirm that the audience can access that environment before comparing convenience or presentation.
Architecture / decision scenario
A sales organization has three requests. Its software team needs a fixed retrieval-and-drafting graph behind a customer portal. Analysts need weekly competitor research followed by a CRM update. An existing Q Business team wants to reuse a successful onboarding conversation as a card-based app.
The portal uses Bedrock Flows: the team tests a draft, publishes a version, and invokes its alias through InvokeFlow. Analysts use Quick Research inside Quick Flows, with configured sources and an Application actions step for the CRM operation. The existing team evaluates Q Apps only after confirming Pro access and the Q Business environment.
A fourth department has no Q Business deployment. It is not offered a new Q Business app because the service is closed to new customers. The organization assesses Apps in Amazon Quick for an interactive application built from a natural-language description, while recording migration questions separately.
Common exam traps
- Treating Bedrock Prompt Flows as a separate product from Bedrock Flows.
- Confusing Quick Flows with Bedrock’s application-invoked flow runtime.
- Assuming a visual builder means business users own the production deployment.
- Offering Q Apps to Lite users or to an unrestricted new Q Business customer.
- Equating a research citation with authorization for a downstream write.
- Assuming similar Quick capabilities guarantee unchanged Q Business API migration.
What to remember
- Select the operating owner and audience before the authoring surface.
- Bedrock versions and aliases provide the application deployment boundary.
- Quick Research and flow steps connect business evidence to actions.
- Q Apps is an existing Q Business environment capability with Pro access.
- Service closure and migration constraints can decide the product choice.
Knowledge check
A software team needs a fixed retrieval-and-drafting graph behind its customer portal. Application code calls it, and changes must reach production only after testing. Which option fits?
InvokeFlowBusiness analysts need weekly cited research on competitors, followed by an update to the matching CRM record, and they want to build and run the sequence themselves. Which option fits?
An existing Amazon Q Business customer wants to turn a successful onboarding conversation into a card-based app for HR staff. Every HR user holds a Q Business Lite subscription. What must change before they can use the app?
A department has never used Amazon Q Business and wants an interactive web app that business users generate from a natural-language description. Which option should it assess?