Chapter 2 of 5 · AWS AIP-C01
Implementation and Integration
Agents and MCP, safeguarded workflows, FM deployment, enterprise integration, FM APIs and developer tools.
- 8topics
- 44 minfast track
- 3diagrams and exhibits
- 26%of the exam
Topics in this chapter
Take them in order, or jump to the one you need.
- 1Agents, Memory, Tools and MCP ServersBuild agents with Strands Agents, AWS Agent Squad and Bedrock AgentCore, give them memory and tools, and host MCP servers on…Diagram6 min
- 2Structured Reasoning, Safeguarded Workflows and Human ReviewUse Step Functions for ReAct loops and chain of thought, add stopping conditions, timeouts, IAM limits and circuit breakers, and…Diagram6 min
- 3Model Coordination, Routing and CascadingCombine specialized FMs, ensembles and model selection, route requests by content or metrics, and cascade from small models to…Diagram6 min
- 4FM Deployment Options and LLM Hosting ChallengesChoose Bedrock on demand, Provisioned Throughput or SageMaker AI endpoints for an FM, and handle LLM needs for GPU memory, tokens…5 min
- 5Enterprise Integration, Secure Access and Hybrid DeploymentConnect FMs to legacy and existing apps with APIs and events, secure access with federation and least privilege, and keep data…5 min
- 6FM APIs, Streaming and Resilient Model CallsCall Bedrock synchronously or via SQS, stream tokens with ConverseStream and API Gateway, manage token limits, and add backoff…6 min
- 7CI/CD, GenAI Gateways and TroubleshootingShip GenAI components with CodePipeline, CodeBuild, tests and rollback, centralize model access in a GenAI gateway, and trace FM…5 min
- 8AI Interfaces, Business Systems and Developer ToolsBuild AI front ends with Amplify, OpenAPI and Bedrock Flows, enhance CRM and document processing with Data Automation, and code…5 min
About this chapter
What this chapter covers
This chapter is 26% of the AWS Certified Generative AI Developer Professional AIP-C01 exam, the second largest domain. It tests whether you can turn a foundation model into a working production system: agents that act, models deployed at the right cost, integrations with existing enterprise systems, and APIs that stay fast and reliable.
The exam has 65 scored questions, either multiple choice or multiple response. Expect scenarios where several options would work, and a phrase such as LEAST operational overhead, MOST cost effective or lowest latency decides the answer.
How the topics fit together
A GenAI application is built in layers, and each layer depends on the one below.
- Agents and workflows give models memory, tools and structured reasoning, with hard safeguards and human review.
- Model coordination decides which model handles which request: specialized models, ensembles, routing and cascades.
- Deployment chooses between Bedrock and SageMaker AI and handles the GPU and token demands of LLMs.
- Enterprise integration connects models to legacy and existing systems securely and within data residency rules.
- FM APIs handle synchronous, asynchronous and streaming calls with resilience built in.
- Delivery and tools cover CI/CD, GenAI gateways, troubleshooting, front ends and developer productivity.
Many answers in this chapter come down to two questions. Is the process fixed (use Step Functions) or open ended (use an agent)? Does the caller need the answer now (synchronous or streaming) or later (queue)?
How to use these pages
Each topic page has the same shape, so you can skim for what you need.
- What it is: the idea in plain words.
- How it works: the mechanism, with diagrams, tables and exam style scenarios.
- Common mistakes: the classic traps.
- What to remember for the exam: the facts to memorize.
- Check yourself: three questions to test your understanding.
Service names change quickly in this area. Learn what each service is for and the requirement that makes it the right choice, because that requirement is what the question will state.
Ready for more practice?
The full AWS AIP-C01 question bank has 827 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by AWS. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.