AWS's Major AI Services, Organized by Use Case [2026 Edition]
- AWS
- AI
- Generative AI
- Amazon Bedrock
- Machine Learning
Introduction
AWS offers a wide range of AI-related services, from services for developing generative AI applications to image recognition, speech processing, and machine learning model development.
However, the roles of these services vary considerably. Some use existing AI models, some train custom models, and others provide specific AI capabilities through APIs.
This article classifies AWS's major AI services by use case and summarizes their features and examples of how they can be used.
*Based on information as of November 2026.
1. Generative AI-Related Services
① Amazon Bedrock
This is a managed service for building applications that use generative AI.
Multiple foundation models, including Anthropic's Claude and Amazon Nova, can be used through APIs. There is no need to develop and train an AI model from scratch, and the burden of infrastructure management can be reduced while incorporating generative AI capabilities into applications.
Main use cases include AI chatbots, text summarization, generating responses to inquiries, and program code generation.
Amazon Nova is a family of AI models developed by AWS and is not itself a service platform like Bedrock.
② Amazon Bedrock AgentCore
This service runs and operates AI agents.
An AI agent is a mechanism in which AI uses external APIs and tools according to the situation and carries out multiple processes. AgentCore provides capabilities such as the agent execution environment, connections to external systems, authentication, and memory management.
For example, it can be used in a setup where AI retrieves inventory information, checks the conditions for placing an order, and asks business systems to perform the necessary operations.
It also supports models and agent frameworks other than those provided through Bedrock.
③ Amazon Bedrock Knowledge Bases
This Amazon Bedrock feature enables generative AI to reference external documents or data when generating responses.
Uses RAG (Retrieval-Augmented Generation) technology to search for relevant information from internal documents and other sources, and passes that content to the AI model. It can also use internal information that the AI model has not been trained on when generating answers.
For example, you can build a chatbot that references internal manuals stored in S3 to answer employees’ questions.
④ Amazon Bedrock Guardrails
This is an Amazon Bedrock feature that provides safety-related controls over generative AI inputs and outputs.
It supports filtering inappropriate content, restricting specific topics, and detecting and masking confidential information such as personal data.
For example, it can be used to control customer-facing chatbots so that personal information is not inadvertently displayed. However, it cannot prevent all incorrect responses or information leaks.
2. Machine Learning Services
⑤ Amazon SageMaker AI
A service that supports the development, training, deployment, and operation of machine learning models.
You can train models using your own data or customize existing models. You can also expose trained models as endpoints for inference and use them from applications.
For example, it can be used to develop models that predict future demand from historical sales data or predict equipment failures from equipment operation data.
Whereas Bedrock is suited to developing generative AI applications that primarily use existing foundation models, SageMaker AI is suited for cases where fine-grained control over model training and the execution environment is required.
⑥ Amazon Personalize
A service that recommends the most suitable products or content for each user.
It uses data such as purchase history, browsing history, and click history to generate recommendations based on user behavior.
For example, it can be used for “recommended products for you” on e-commerce sites or to display recommended content on video streaming services.
3. Image and Document Analysis Services
⑦ Amazon Rekognition
A service that uses AI to analyze the content of images and videos.
It supports detecting objects and people in images, analyzing facial features, detecting inappropriate content, and more. It can also detect people and objects in videos.
For example, it can be used to automatically classify uploaded images based on their content or detect specific objects within videos.
⑧ Amazon Textract
A service that extracts text and data from PDFs and scanned documents.
In addition to extracting text through OCR (Optical Character Recognition), it analyzes the structure of tables and forms and can obtain relationships between field names and values.
For example, it can be used to extract client names, billing amounts, billing dates, and other information from invoices and register it in a business system.
4. Text and Audio Processing Services
⑨ Amazon Comprehend
A service that analyzes text content using natural language processing.
It supports extracting keywords and named entities from text, sentiment analysis, language identification, text classification, and more.
For example, it can be used to analyze customer survey responses and classify positive and negative opinions.
⑩ Amazon Transcribe
A service that converts audio into text.
It supports not only prerecorded audio files but also real-time audio streaming. It also includes features such as speaker identification and custom vocabularies for handling specialized terminology.
For example, it can be used to prepare meeting minutes, transcribe call-center conversations, and create video subtitles.
⑪ Amazon Polly
A service that converts text into speech.
It supports multiple languages and voices and can generate spoken audio that reads text aloud. Depending on the type of voice, features such as neural speech synthesis may also be available.
For example, it can be used to read website content aloud, provide audio guidance, and power navigation systems.
⑫ Amazon Translate
A service that automatically translates text into different languages.
It uses neural machine translation to perform translation via the API from applications. Custom glossaries for adjusting terminology translations are also available.
For example, it can be used for multilingual websites and translating inquiries from overseas customers.
⑬ Amazon Lex
A service for building conversational interfaces using voice and text.
It determines the user's intent and required information from the user's utterance and proceeds according to a predefined conversational flow.
For example, it can be used for hotel reservation handling, checking delivery status, and automated call-center response systems.
5. Development and Operations Support Services
⑭ Amazon Q Developer
A generative AI assistant that supports AWS development and operations.
It assists with questions about using AWS services, code generation, code explanation, security analysis, troubleshooting, and more.
For example, it can be used to help create CloudFormation templates and investigate the causes of errors that occur in AWS environments.
The Amazon Q Developer IDE plugin is scheduled to end support on April 30, 2027, and AWS is directing users to migrate to Kiro.
⑮ Kiro
A development tool that uses AI to assist with software development.
Based on requirements expressed in natural language, it organizes requirements definition, design, and implementation tasks, and assists with creating code and tests. It particularly emphasizes specification-driven development, in which implementation begins after the specifications have been clarified.
For example, when developing a new web application, it can support the entire workflow from organizing requirements through implementing the code.
⑯ Amazon Quick Suite
A service that uses AI to support enterprise information search, analysis, and business process automation.
It can search and analyze information across internal documents, databases, business applications, and other sources, enabling work to proceed based on the results.
For example, it can be used to analyze internal sales data and create reports, or to integrate with multiple business systems and automate routine tasks.
It was introduced as an advanced version of Amazon Q Business, and the existing Q Business can also continue to be used.
6. Differences between similar services
Some AWS AI services have similar purposes. Below are services that are particularly easy to confuse.
| Services being compared | Key differences|
|---|---|
| Bedrock / SageMaker AI | A foundation for building generative AI applications / A platform for developing, training, and operating models |
| Rekognition / Textract | Analyzes the content of images and videos / Extracts text and structured data from documents |
| Transcribe / Polly | Converts speech to text / Converts text to speech |
| Comprehend / Bedrock | Analyzes text, including classification and information extraction / Flexible text processing using generative AI models |
| Lex / Bedrock | Conversational processing based on predefined conversation flows / Flexible conversational generation using foundation models |
| Bedrock / Q Developer | A foundation for building generative AI applications / An AI assistant that supports AWS development and operations |
| Bedrock / Quick Suite | Build custom generative AI applications / Use enterprise information search, analysis, and business support features |
These are not necessarily mutually exclusive.For example, you can also build conversational systems that integrate Lex with Bedrock, or systems that use Bedrock to analyze document data extracted by Textract.
7. AWS Services That Integrate with AI Services
When actually building a system that uses AI, it is common to combine AI services with existing AWS services rather than use an AI service on its own.
| AWS Service | Primary Role in AI Systems |
|---|---|
| Amazon S3 | Stores documents, images, audio, and other data processed by AI |
| AWS Lambda | Calls AI service APIs and processes results |
| Amazon API Gateway | Exposes AI functionality to external applications as an API |
| AWS Step Functions | Controls the sequence and branching of multiple AI and business processes |
| AWS IAM | Manages access permissions for AI services and data |
| Amazon CloudWatch | Monitors processing status, logs, errors, and other conditions |
For example, a mechanism for extracting information from PDFs could be configured to store the PDFs in S3, invoke Textract from Lambda, and register the extracted results in a database.
When using AI services, you also need to consider fees based on the number of processing operations and data volume, available AWS Regions, and the confidentiality of the data being handled.
Summary
AWS AI-related services play very different roles depending on their use.
If you want to develop an application that uses generative AI, Amazon Bedrock is the primary option. If you want to develop and operate your own machine learning models, Amazon SageMaker AI is the primary option.
On the other hand, when the objective is clear, such as image recognition, document analysis, transcription, or translation, you can use an AI service specialized for that purpose. For development support, there are options such as Amazon Q Developer and Kiro, while Amazon Quick Suite is available for leveraging information within an organization.
What is important is not to memorize all the services from the outset, but to be able to determine which service is best suited to the functionality you want to achieve.
References
Go to list of users who liked
Go to list of comments
Register as a new user and use Qiita more conveniently
- You get articles that match your needs
- You can efficiently read back useful information
- You can use dark theme
