A fast pass over the managed AI services: one line each, and the pairs that look interchangeable but are not.
Services at a glance
| Service | What it does | Reach for it when |
|---|---|---|
| Amazon Polly | Text to speech | Content needs to be spoken aloud |
| Amazon Transcribe | Speech to text | Audio needs to become a transcript |
| Amazon Translate | Language translation | Text needs to move between languages |
| Amazon Comprehend | Text analysis: sentiment, entities, PII detection | You need to understand what a body of text contains, not generate new text |
| Amazon Lex | Conversational interfaces | You are building a chatbot or voice bot with defined intents |
| Amazon Rekognition | Image and video analysis | You need to detect objects, faces, text, or moderate content in visual media |
| Amazon Textract | Document structure extraction | You need forms, tables, and key-value pairs out of a scanned or digital document |
| Amazon Personalize | Recommendations | You need to rank or suggest items for a user based on behaviour |
| Amazon Quick | A managed assistant over company data | You need connectors, permissions, and citations with no code written |
Some names have come off the in-scope service list, so a familiar answer can now be the wrong one, and the names that came off it are worth learning as a set. Amazon Kendra went into maintenance mode on 30 June 2026 and closed to new customers on 30 July, so existing indexes keep running and a new retrieval build resolves to Amazon Bedrock Knowledge Bases rather than a Kendra index. Amazon Forecast and Amazon Fraud Detector are closed to new customers and are not listed either, so a forecasting or fraud-detection scenario resolves to Amazon SageMaker AI; the use case is still fair game even though the named service is not. Amazon Q Developer is not on the list, and it is the easiest of them to answer with by reflex, because it does genuine developer work from the IDE, the CLI, and the console. Kiro is the developer tool that is listed, so a code scenario resolves there.
The GenAI build surface at a glance
The table above is the packaged half of the list, one fixed job per service. The generative half is where you build the job nobody has packaged yet, and the blocks stack in a predictable order.
| Service | What it does | Reach for it when |
|---|---|---|
| Amazon Bedrock | A managed API over foundation models from several providers, with no infrastructure to run | An application needs to call a hosted model and pay by the token |
| Amazon Nova | AWS’s own foundation-model family, served on Bedrock and tiered from fast and cheap up to the flagship | You want a first-party model and want to trade capability against cost inside one family |
| Amazon SageMaker AI | Build, train, tune, and host your own models on endpoints you operate | No purpose-built service fits and the model has to be yours |
| Amazon SageMaker JumpStart | A hub of pre-trained open-source and third-party models you deploy onto SageMaker endpoints | You want a ready-made model, but the weights have to sit on infrastructure you chose |
| Amazon Bedrock AgentCore | Managed runtime, memory, gateway, and identity for agents in production | An agent needs somewhere to run, remember, and authenticate, and you are not building that yourself |
| Strands Agents | An open-source framework for writing agents in code: model, tools, loop | The agent’s behaviour is yours to write rather than configure |
| Kiro | The AWS developer assistant in the IDE | Someone is writing software and wants help writing it |
| AWS Transform | Agentic modernisation of legacy workloads | Old code or infrastructure has to move forward and agents can do the repetitive part |
The Bedrock surface at a glance
Bedrock is one service with several named features, and they get named. Grounding answers in your own documents is the one that turns up most often.
| Feature | What it does | Reach for it when |
|---|---|---|
| Amazon Bedrock Knowledge Bases | Managed RAG: ingest, chunk, embed, store, and retrieve, then hand the passages to the model | Answers have to come from your own documents, and those documents change |
| Amazon Bedrock Agents | An agent configured in the console and run by Bedrock: it plans, calls tools, and acts in other systems, with no runtime of yours involved | The job ends in something happening rather than in text coming back, and configuring the agent beats writing one |
| Amazon Bedrock Guardrails | Content filters, denied topics, PII redaction, and grounding checks applied around the model call | Something has to be blocked whether or not the model cooperates |
| Amazon Bedrock Prompt Management | Prompts stored as versioned resources with input variables | Wording changes often enough that a change has to be traceable and reversible |
| Amazon Bedrock Model Evaluation | Automatic, human, and side-by-side comparison jobs scoring model output | You need evidence for a model choice rather than an impression |
Responsible-AI documentation on these services
AWS publishes an AI Service Card for several of the services above, Rekognition face matching and the Titan and Nova models among them. Each card states intended use, fairness and performance considerations, and deployment guidance. It is AWS’s document about an AWS service. Your team’s equivalent for a model you built is an Amazon SageMaker Model Card, one of the standard tools for identifying transparent and explainable models. Both serve transparency, and they differ in who wrote them and about what.
Two of the purpose-built services do responsible-AI work themselves. Rekognition moderates images and video, and Comprehend detects toxicity in text. Both cover the safety dimension for a team with no foundation model anywhere in the design.
Decision rules
- If the job is turning text into audio, that is Polly; if it is turning audio into text, that is Transcribe. The direction of the arrow decides which one.
- If the task is understanding existing text (sentiment, entities, PII), reach for Comprehend rather than asking a foundation model to do analysis a purpose-built service already does more cheaply.
- If you are building a defined-intent chatbot or voice interface, use Lex; an open-ended assistant over your own documents is Amazon Quick, not a Lex bot.
- If the input is an image or video and the question is “what is in this”, use Rekognition; if the input is a document and the question is “what is its structure”, use Textract.
- If you need a retrieval layer feeding your own application, that is Amazon Bedrock Knowledge Bases; if you need a ready assistant with a UI, connectors, and permissions, that is Amazon Quick. Kendra held the first slot until it closed to new customers, so only an account that already runs an index still has it as an option.
- If the ask is “recommend items to this user”, that is Personalize, not a general-purpose model prompted with behavioural data.
- If a request is about writing code, that is Kiro; if it is about answering questions from company documents, that is Amazon Quick. Amazon Q Developer covers similar developer ground from the IDE, the CLI, and the console, which is why it gets named in the same breath as both, and it is not on the list.
- If a purpose-built AI service already solves the problem, prefer it over building the same capability on a foundation model; it is usually cheaper, faster, and more consistent for a narrow task.
- Capability descriptions cluster on six managed AI/ML services: Amazon SageMaker AI, Amazon Transcribe, Amazon Translate, Amazon Comprehend, Amazon Lex, and Amazon Polly. A scenario about what a managed service can do is nearly always one of those six.
- Sort Amazon Quick, Kiro, and Amazon Bedrock by who is asking. A business user with a question about company data gets Amazon Quick; a developer writing software gets Kiro; an application that needs a model behind its own code gets Bedrock.
- Bedrock or SageMaker AI comes down to who operates the endpoint. Bedrock is an API call to a model AWS runs and bills by the token; SageMaker AI is a model you deploy, size, and pay for while it is up.
- Amazon SageMaker JumpStart rather than Bedrock when the weights have to land on infrastructure you chose: your instance type, your VPC, your account.
- If answers have to come from the company’s own current documents, that is Amazon Bedrock Knowledge Bases and not fine-tuning. Retrieval keeps pace with documents that change; training does not.
- If the model has to take an action in another system, that is an agent, not a longer prompt.
- If a wording change has to be rolled back, that is Amazon Bedrock Prompt Management, where prompts are versioned resources with input variables.
- If a scenario asks what AWS says the limits of a managed AI service are, that is the AI Service Card; if it asks what the limits of your own model are, that is a SageMaker Model Card.
Traps
- Assuming Comprehend generates text. It analyses text that already exists; it does not write anything new.
- Confusing Rekognition and Textract because both process images. Rekognition answers “what is depicted”; Textract answers “what is the structure of this document” (fields, tables, key-value pairs).
- Treating Kendra as a chatbot. It is the retrieval layer other things are built on top of, not a conversational surface in itself.
- Picking Kendra for a new build. It went into maintenance mode on 30 June 2026 and closed to new customers a month later; existing indexes keep running, but a new retrieval build goes to a Bedrock knowledge base.
- Swapping the assistants in a sentence about “the AWS assistant”. Amazon Quick answers from company data with citations and no code written. Kiro is the listed developer tool and works where software gets written. Amazon Q Developer does comparable developer work from the IDE, the CLI, and the console, and is not on the list. A question about company documents is never answered by either developer tool.
- Reaching for a foundation model to do sentiment analysis or entity extraction when Comprehend already does it as a managed, purpose-built call.
- Picking Personalize for a one-off recommendation with no behavioural history to train on; it needs interaction data to be worth using over a simpler rule.
- Forgetting that Lex handles defined intents and slots; an open-ended, document-grounded conversation is a different architecture, not a Lex bot with more intents bolted on.
- Treating Strands Agents and Amazon Bedrock AgentCore as competing choices. Strands Agents is the framework the agent is written in; AgentCore is the managed runtime it can be deployed onto. Picking one does not rule out the other, and Amazon Bedrock Agents is a third shape again: an agent you configure rather than write, run by Bedrock rather than deployed onto a runtime of yours.
- Reading Bedrock Guardrails and a prompt instruction as the same control. A guardrail is a filter around the model call and applies whether the model cooperates or not; “never discuss competitors” inside a prompt is a request the model can ignore.
- Confusing Amazon Bedrock Model Evaluation with SageMaker Clarify. Model Evaluation scores what a model produced; Clarify measures bias and feature attribution.
- Answering a forecasting or fraud-detection scenario with Amazon Forecast or Amazon Fraud Detector. Both are closed to new customers and off the in-scope list, so that work goes to SageMaker AI.
Say it in one line
- Polly turns text into speech; Transcribe turns speech into text. Pick by which way the conversion runs.
- Comprehend analyses text that already exists: sentiment, entities, PII. It does not generate anything.
- Lex builds defined-intent chatbots and voice bots from intents and slots; an open-ended assistant is a different build.
- Rekognition reads images and video; Textract reads document structure, forms, and tables.
- Kendra was the enterprise-search retrieval layer and is now maintenance-only and closed to new customers; Amazon Bedrock Knowledge Bases is where a new retrieval build goes, and Amazon Quick is the ready assistant with connectors, permissions, and citations.
- Personalize ranks and recommends from behavioural data.
- Amazon Quick answers questions from company data with citations and no code; Kiro is the listed tool for people writing software. Amazon Q Developer does the same kind of developer work as Kiro and is not on the list, so a code scenario resolves to Kiro.
- Prefer a purpose-built AI service over a foundation model when one already exists for the task.
- Bedrock is a model AWS operates and bills by the token; SageMaker AI is a model you operate on endpoints you size and pay for; JumpStart is how a ready-made model gets onto them.
- Knowledge Bases grounds answers in documents that change; fine-tuning does not keep up with them.
- Guardrails filter around the model call; a prompt instruction is a request the model can ignore.
- An AI Service Card is AWS’s account of an AWS service; a SageMaker Model Card is your team’s account of your model.