Skip to main content
Welcome. This lesson reviews common generative AI use cases and real-world applications, focusing on practical examples, the model families that power them, and the platforms that host them. This material is important and likely to appear on the exam — know both the “what” (use case) and the “how” (model family and platform).
Focus on the distinction between use cases (what you want to accomplish) and model families or platforms (how you implement it). Exam questions often test your ability to map a business need to the appropriate model type and cloud service.

​
Main use cases

Generative AI models are adaptable across many domains. Below are the primary categories you should recognize, with concrete examples and common objectives for each.
  • Text generation and adaptation — producing or rewriting content to match an objective, audience, or readability level (e.g., marketing copy, technical docs, student-level explanations).
  • Summarization and information extraction — condensing long documents, meeting transcripts, or reports while preserving key facts and named entities.
  • Code generation and developer assistance — creating boilerplate, suggesting completions, or generating structured artifacts such as YAML and JSON for infrastructure or CI/CD.
  • Audio and visual creation — generating images, video, 3D content, voice synthesis, and other creative media for marketing, entertainment, or simulation.
  • Personalization and dynamic content — customizing product descriptions, recommendations, or conversational responses for user segments or contexts.
  • Document understanding and compliance — extracting clauses, identifying regulatory risks, and automating contract review at scale.
Text generation is often more than just “writing.” It includes adapting technical content to different audiences, rewriting for clarity or tone, and extracting categorized summaries or key points from long documents.
A presentation slide titled "Text Generation – Adapting Content for Different Audiences." It shows two numbered panels: "01 Adapting technical content" with a monitor/gear icon and "02 Rewriting text for varying levels of expertise" with a phone-and-pencil icon.

​
Examples and quick reference

​
Code generation

Code-focused generative tools accelerate development by automating routine tasks, suggesting completions, and generating snippets or configuration files. Popular examples include GitHub Copilot, Google Gemini, AWS CodeWhisperer, Tabnine, and Cursor AI. Common uses:
  • Producing example code and templates for common libraries and frameworks.
  • Auto-completing functions and method bodies, including docstrings and tests.
  • Generating or validating structured files such as YAML and JSON for deployment manifests and APIs.
Tool selection often depends on IDE integration, supported languages, security/data residency needs, and cost.
A presentation slide titled "Generative AI in Code Generation" showing two numbered cards: 01 ("Automating routine coding tasks" with a gear icon) and 02 ("Code completions and suggestions" with a code-file icon). The design uses a dark blue background with colored accents.

​
Cloud services and platforms

Managed services reduce operational overhead and provide production-ready APIs for generative AI. Key AWS services and when to consider them: These services simplify integration of generative capabilities into applications while handling infrastructure, scaling, security, and monitoring.
A slide titled "Generative AI — AWS Services" with three numbered columns. The columns list 01 Amazon Bedrock and Amazon Titan, 02 Amazon Q Developer (formerly CodeWhisperer), and 03 Amazon SageMaker, each shown with a simple icon.

​
Information extraction, personalization, and industry use

Generative models are particularly valuable for extracting facts and structured data from unstructured sources:
  • Legal: Identify clauses, extract obligations, or flag risky terms in contracts.
  • Finance: Summarize earnings reports, extract KPIs, and detect anomalies.
  • Healthcare: De-identify notes, extract clinical concepts, and generate discharge summaries.
  • Marketing and e-commerce: Generate personalized product descriptions, dynamic ad copy, and customer-facing FAQs.
These capabilities enable automation at scale, improved customer personalization, and faster insights from large text corpora.

​
Architectures behind generative AI

Different model families are suited to different data types and outputs. Recognize the main architectures and their trade-offs: Transformers form the backbone of most modern LLMs (e.g., GPT, Gemini). Choose the architecture based on the output format (text vs. image), control needs, and compute constraints.
A dark-themed slide titled "Architectures Behind Generative AI — GAN, VAE, Transformers." It shows three numbered panels labeled 01 Generative Adversarial Network (GAN), 02 Variational Autoencoder (VAE), and 03 Transformers, each with a simple white icon.

​
Summary

Generative AI covers a wide range of practical applications: content generation, summarization, code assistance, and media creation. When designing solutions:
  • Start with the use case and required outputs (text, code, image, structured data).
  • Map the use case to an appropriate model family (transformer, GAN, VAE).
  • Choose a platform that meets operational needs (Bedrock, SageMaker, or a specialized tool).
  • Consider constraints: data availability, latency, cost, and security/compliance.
Understanding these trade-offs and common services will help you design robust generative AI solutions and answer exam questions that require mapping requirements to models and platforms.

Watch Video