4.2.5. The Amazon Q Family: Business and Developer

💡 First Principle: Amazon Q is AWS's family of managed generative AI assistants — you bring the data and the identity, AWS runs the model and the retrieval. Q Business targets enterprise knowledge work; Q Developer targets the software lifecycle. The exam tests which member fits a scenario and how each respects enterprise identity and permissions.

The Amazon Q family at a glance:

Status: Amazon Q Business has been closed to new customers since July 30, 2026; existing customers keep it with bug fixes and security updates. AWS recommends Amazon Quick for new deployments — its knowledge bases enforce document-level ACLs for S3, SharePoint, Confluence Cloud, and Google Drive, and it signs users in through IAM Identity Center. Both remain in scope for this exam.

Amazon Q Business — identity and access:

• Connects to enterprise data sources and honors each document's existing access controls, so a user sees only content their permissions already allow. It integrates with IAM Identity Center for user identity and group membership.
• Plugins let users trigger actions in external systems such as a ticketing tool. An administrator scopes who may run which plugin actions through Identity Center groups — never by sharing one set of credentials.
• Q Business retrieves from indexed content; it does not train a new foundation model on customer documents by default.

Amazon Q Developer — the software lifecycle:

• Works inside the IDE (VS Code, JetBrains) to generate, explain, and refactor code, and troubleshoots errors in the AWS console.
• Code transformation automates large language-version upgrades such as Java 8 to Java 17, updating dependencies and constructs across the codebase.
• Customizations index an organization's private code repositories so suggestions follow internal libraries and patterns, with access scoped through IAM Identity Center. This is the governed alternative to pasting internal code into prompts.
• Status: new Q Developer sign-ups have been blocked since May 15, 2026, and the Q Developer IDE plugins and paid subscriptions (code transformation included) reach end of support on April 30, 2027. AWS points IDE users to Kiro and Java upgrades to AWS Transform custom. Q Developer in the AWS Management Console and in chat apps (Slack, Microsoft Teams) is not affected, and Q Developer remains in scope for this exam.

⚠️ Exam Trap: Don't confuse the two. Answering employee questions from SharePoint or Confluence with per-user permissions points to Q Business (or, for a new deployment, its successor Amazon Quick). Upgrading a legacy codebase or tailoring code suggestions to internal repos points to Q Developer (a new Java upgrade project today goes to AWS Transform custom). A base Bedrock model has no connection to enterprise content or its permission model, so it cannot enforce per-user access on its own.

Kiro: spec-driven agentic development

Kiro is AWS's AI-powered development environment. One agent works across an IDE, a CLI, and web and mobile apps, and its configuration lives in the repository's .kiro/ folder. Where Q Developer is centered on suggestions, transformation, and troubleshooting, Kiro is built around planning the work before the agent writes code.

• Spec mode turns a feature request into three files under .kiro/specs: requirements.md (user stories and acceptance criteria in EARS notation, such as WHEN a user submits a prompt over the token limit THE SYSTEM SHALL return a validation error), design.md (architecture and sequence diagrams), and tasks.md (discrete, trackable tasks). Vibe mode is the chat-based alternative with no formal plan.
• Steering files in .kiro/steering (defaults product.md, tech.md, structure.md) give Kiro standing project knowledge. They can load always, when matching files are in play, on request, or automatically. Workspace steering overrides global steering in ~/.kiro/steering.
• Hooks run a shell command or an agent prompt when events happen, such as FileSave, PromptSubmit, or PreToolUse. A PreToolUse hook can block a risky operation.
• MCP servers connect Kiro to external tools and data.

⚠️ Exam Trap: "Produce reviewable requirements, a design, and a task list before code is written" points to Kiro specs, not Q Developer code transformation (which upgrades existing code) or Q Developer customizations (which tailor suggestions to private repositories). "Apply our standards in every session" is steering; "run tests whenever a file is saved" is a hook.

Reflection Question: An enterprise wants employees to query internal wikis and ticketing data through a chat assistant, seeing only what their role permits, and separately wants developers to upgrade a Java 8 service to Java 17. Which Amazon Q members address each need, and how is per-user access enforced in the first case?