Skip to content
certsim
About

Services

What each service is for, and what it gets confused with.

36 results · content in English

Amazon Bedrock

In scope

Managed access to foundation models and tools for building generative AI applications.

  • An insurance claims chatbot calls a foundation model in Amazon Bedrock and pays per token with no commitment.
  • A retailer classifies millions of product reviews overnight with a Bedrock foundation model; results are needed the next day.

Not to be confused with: Amazon SageMaker AI

Amazon Bedrock AgentCore

In scope

Managed runtime and supporting services for deploying and operating agents.

  • A platform team must deploy and operate agents built with any framework, with session isolation, identity, and observability.

Not to be confused with: Strands Agents · Amazon Bedrock

Amazon Bedrock Knowledge Bases

Out of scope

Managed retrieval and grounding workflows over connected enterprise data.

  • Answer employee questions using the company's private documents through a managed RAG workflow.
  • Show users which source documents support each generated answer.

Not to be confused with: Fine-tuning

Amazon Bedrock Guardrails

Out of scope

Configurable controls for harmful content, sensitive information and other application policies.

  • Block responses about denied topics and filter harmful content in inputs and outputs.

Not to be confused with: AWS IAM · Amazon Macie

Amazon SageMaker AI

In scope

Build, train and deploy custom machine learning models.

  • A clinic's custom SageMaker AI model receives small requests a few times per hour, must not incur idle cost, and can tolerate cold starts.
  • A genomics lab sends very large files to a custom SageMaker AI model; each analysis runs a long time and results are retrieved later from Amazon S3.

Not to be confused with: Amazon Bedrock

Amazon SageMaker JumpStart

In scope

Discover pretrained models and deploy or customize supported models.

  • Deploy a publicly available open-weight model to an endpoint in the company's account on an instance type the team selects.
  • A team wants to choose an open-weight model from a catalog, fine-tune it with prebuilt workflows, and deploy it to an endpoint on instance types it selects.

Not to be confused with: Amazon Bedrock

Amazon SageMaker Clarify

Out of scope

Measure bias and explain model predictions with feature attributions.

  • Before training, measure whether the loan dataset under-represents applicants in a specific age group.
  • Show which features contributed most to an individual applicant's loan decision.

Not to be confused with: Amazon SageMaker Model Monitor

Amazon SageMaker Model Monitor

Out of scope

Monitor production data and model quality, including supported bias and feature-attribution drift checks.

  • Detect drift in production data and model quality against a baseline
  • Receive an alert when bias metrics on live predictions move outside an allowed range.

Not to be confused with: Amazon CloudWatch

Amazon SageMaker Ground Truth

Out of scope

Create training labels using human workforces and supported automated labeling workflows.

  • Have a private workforce draw bounding boxes on product images to create training labels

Not to be confused with: Amazon Augmented AI

Amazon SageMaker Feature Store

Out of scope

Store and share model features for training and online inference.

  • Store engineered features once and serve the same values for training and low-latency inference

Not to be confused with: Amazon S3

Amazon SageMaker Model Cards

Out of scope

Document intended uses, risks, training and evaluation information for a model.

  • Document the model's intended use, risk rating, and evaluation results for internal auditors.
  • Document the intended uses, risk rating, and evaluation results of a credit model that the company trained in SageMaker AI, for internal auditors.

Not to be confused with: AWS AI Service Cards · AWS Artifact

Amazon Comprehend

In scope

Extract insights from text, such as sentiment, entities and supported PII detection.

  • Determine whether product reviews are positive, negative, neutral, or mixed.

Not to be confused with: Amazon Textract · Amazon Macie

Amazon Lex

In scope

Build conversational voice and text interfaces using intents and slots.

  • Build a text and voice bot that collects order details from customers.

Not to be confused with: Amazon Polly · Amazon Transcribe

Amazon Polly

In scope

Convert text into spoken audio with voice and pronunciation controls.

  • Read news articles aloud in a natural-sounding voice.
  • Synthesized audio for a legal disclaimer needs an added pause and a slower speaking rate.

Not to be confused with: Amazon Transcribe

Amazon Rekognition

In scope

Analyze images and video for objects, text and other supported visual features.

  • Read license plate numbers from traffic camera images.
  • Image analysis must detect team logos on player jerseys, which general label detection does not recognize.

Not to be confused with: Amazon Textract

Amazon Textract

In scope

Extract document text, forms and tables from scanned documents.

  • Extract key-value pairs and tables from scanned tax forms.

Not to be confused with: Amazon Comprehend · Amazon Rekognition

Amazon Transcribe

In scope

Convert speech to text, with features such as custom vocabulary.

  • Convert recorded customer voicemails into text.
  • Speech-to-text output keeps misspelling a short list of the company's brand names.

Not to be confused with: Amazon Polly

Amazon Translate

In scope

Translate text across supported languages; custom terminology controls selected terms.

  • A product name must stay untranslated in French output instead of being translated literally.

Not to be confused with: Amazon Transcribe

Amazon Personalize

In scope

Create personalized recommendations from user interaction and item data.

  • Create personalized recommendations from user interaction and item data.

Not to be confused with: Amazon Comprehend

Amazon Nova

In scope

AWS foundation model family with capabilities that differ by model and modality.

  • AWS foundation model family with capabilities that differ by model and modality.

Not to be confused with: Amazon Bedrock

Amazon Quick

In scope

AI-assisted analysis and business insights, including dashboard-oriented workflows.

  • Business users want AI-assisted dashboards and data analysis without hosting any models
  • Business analysts ask natural-language questions about sales data and get charts and dashboards.

Not to be confused with: Amazon SageMaker AI

Kiro

In scope

Agentic development tooling for software creation and development workflows.

  • Developers want an agentic IDE that writes requirements, a design, and tasks before generating code.

Not to be confused with: AWS Transform · Strands Agents

Strands Agents

In scope

An SDK for building agents and composing agent workflows.

  • Developers want an open-source SDK to define an agent in code with a model, tools, and a prompt.

Not to be confused with: Amazon Bedrock AgentCore

AWS Transform

In scope

Agentic assistance for modernization and transformation of existing workloads.

  • An IT team wants AI agents to plan and execute the modernization of mainframe and VMware workloads.

Not to be confused with: Kiro

AWS CloudTrail

In scope

Record AWS API activity for auditing who performed an action and when.

  • Record which IAM principal deleted a Bedrock knowledge base and when.

Not to be confused with: Amazon CloudWatch · AWS Config

Amazon CloudWatch

In scope

Collect operational metrics and logs and configure alarms.

  • Monitoring: alert the operations team when invocation errors or latency rise above a threshold.

Not to be confused with: AWS CloudTrail

AWS Config

In scope

Record and evaluate AWS resource configurations against rules.

  • Continuously evaluate whether SageMaker AI endpoint configurations use AWS KMS keys, and flag noncompliant resources.

Not to be confused with: AWS CloudTrail · AWS Artifact

AWS Artifact

In scope

Access AWS compliance reports and agreements.

  • Download AWS SOC reports and ISO certifications to show auditors that the underlying AWS services meet compliance standards.
  • Download AWS third-party audit reports for a compliance review

Not to be confused with: AWS Config · Amazon SageMaker Model Cards

Amazon Inspector

In scope

Find supported software vulnerabilities and unintended network exposure.

  • Scan EC2 instances and container images that host a self-managed model for software vulnerabilities
  • Continuously scan Lambda functions that call the model for known software vulnerabilities.

Not to be confused with: Amazon Macie

AWS Key Management Service (AWS KMS)

In scope

Create and control cryptographic keys used to protect data.

  • Create, rotate, and control access to the encryption keys that protect training data

Not to be confused with: AWS Secrets Manager

Amazon Macie

In scope

Discover sensitive data in Amazon S3 and assess related data-security risks.

  • Discover PII in S3 buckets that hold model training data
  • Data discovery: identify which S3 buckets that hold training data contain PII.

Not to be confused with: Amazon Comprehend · Amazon Inspector

AWS Identity and Access Management (IAM)

In scope

Control which principals can perform actions on AWS resources.

  • Control which principals can perform actions on AWS resources.

Not to be confused with: Amazon Bedrock Guardrails

Amazon S3

In scope

Object storage for training datasets, model artifacts and batch inputs or outputs.

  • Store billions of vectors at the lowest cost for a workload that is queried infrequently.
  • Retention: automatically delete claim summaries in Amazon S3 when the approved retention period ends.

Not to be confused with: Amazon SageMaker Feature Store

AWS Glue

In scope

Serverless data integration, cataloging and ETL capabilities.

  • Data engineers need serverless ETL jobs to extract, transform, and load raw data into a data lake

Not to be confused with: Amazon SageMaker AI

AWS Trusted Advisor

In scope

Provide recommendations across supported AWS best-practice checks.

  • Get recommendations about open S3 bucket permissions, idle resources, and service limits.

Not to be confused with: AWS Config · Amazon Inspector