AIP-C01 Domain 2: Implementation and Integration (26%)
Domain 2 carries 26% of the AIP-C01 exam. It tests whether you can build agentic AI solutions with tool integrations, deploy models, connect GenAI to enterprise systems, integrate FM APIs with streaming, retries and routing, and use development tools.
Weight on the exam: 26%. It is one of the five domains of the AWS Certified Generative AI Developer – Professional AIP-C01 exam. Below is a summary of its tasks and skills according to the official AWS guide.
Tasks and skills tested
Task 2.1: Implement agentic AI solutions and tool integrations
- Build autonomous systems with memory and state management, using Strands Agents, AWS Agent Squad and MCP for agent-tool interactions.
- Implement structured reasoning patterns such as ReAct and chain-of-thought with Step Functions.
- Safeguard workflows with stopping conditions, Lambda timeouts, IAM resource boundaries and circuit breakers.
- Coordinate several models: specialized FMs, custom aggregation logic for ensembles and model selection frameworks.
- Add human expertise with Step Functions review and approval flows and API Gateway feedback collection.
- Integrate tools reliably with standardized function definitions, error handling and parameter validation in Lambda.
- Extend FMs with MCP servers: stateless servers on Lambda for lightweight tools and servers on Amazon ECS for complex tools.
Task 2.2: Implement model deployment strategies
- Deploy FMs to match application needs: Lambda for on-demand invocation, Bedrock provisioned throughput and SageMaker AI endpoints for hybrid solutions.
- Address what is different about LLM deployment: container patterns tuned for memory, GPU use and token processing, and specialized model loading.
- Balance performance and resources by choosing smaller pre-trained models for specific tasks and API-based model cascading for routine queries.
Task 2.3: Design and implement enterprise integration architectures
- Connect FMs to enterprise environments with API integrations to legacy systems, event-driven loose coupling and data synchronization patterns.
- Add GenAI to existing applications using API Gateway microservice integrations, Lambda webhook handlers and Amazon EventBridge.
- Apply secure access frameworks: identity federation, role-based access control and least privilege API access to FMs.
- Meet data compliance across jurisdictions with AWS Outposts, AWS Wavelength and secure routing between cloud and on-premises resources.
- Build CI/CD pipelines and GenAI gateways with CodePipeline, CodeBuild, automated tests, security scans, rollback support and centralized abstraction layers.
Task 2.4: Implement FM API integrations
- Create flexible model interactions: Bedrock APIs for synchronous requests, language-specific SDKs and Amazon SQS for asynchronous processing, and API Gateway with request validation.
- Deliver real-time responses with Bedrock streaming APIs, WebSockets or server-sent events and chunked transfer encoding.
- Make FM systems resilient with SDK exponential backoff, API Gateway rate limiting, fallback mechanisms and AWS X-Ray tracing.
- Route requests to the right model with static routing in code, Step Functions content-based routing, metrics-based routing and API Gateway request transformations.
Task 2.5: Implement application integration patterns and development tools
- Build FM API interfaces that handle streaming responses, token limits and retries for model timeouts.
- Speed up adoption with AWS Amplify UI components, OpenAPI specifications and Bedrock Prompt Flows as a no-code builder.
- Enhance business systems with Lambda CRM enhancements, Step Functions document processing and Bedrock Data Automation.
- Improve developer productivity with Amazon Q Developer for code generation, refactoring, testing and optimization.
- Build advanced applications with Strands Agents, AWS Agent Squad, Step Functions agent patterns and Bedrock prompt chaining.
- Troubleshoot faster with CloudWatch Logs Insights over prompts and responses, X-Ray traces of FM API calls and Amazon Q Developer error pattern recognition.
Services and tools to master
- Amazon Bedrock
- Amazon Bedrock AgentCore
- Strands Agents
- AWS Agent Squad
- Model Context Protocol (MCP)
- AWS Step Functions
- AWS Lambda
- Amazon ECS
- Amazon API Gateway
- Amazon EventBridge
- Amazon SQS
- AWS CodePipeline
- AWS CodeBuild
- AWS Outposts
- AWS Wavelength
- Amazon SageMaker AI
- AWS Amplify
- Amazon Bedrock Data Automation
- Amazon Q Developer
- AWS X-Ray
Key facts for domain 2
Short summaries; each links to the study topic with the full note and its official source.
- Provisioned Throughput — Provisioned Throughput reserves model units (MUs) at a fixed hourly price, with no commitment or a 1-month or 6-month term (longer = cheaper, and you cannot delete it before the term ends).
- Inference profiles and Provisioned Throughput — Inference profiles (cross-Region inference) currently do not support Provisioned Throughput.
- Batch inference — Batch inference processes many prompts asynchronously: you upload input files to S3, submit a job and read the results from S3.
- Streaming APIs — Use ConverseStream, InvokeModelWithResponseStream or RetrieveAndGenerateStream to send tokens to the user as they are generated instead of waiting for the full response.
- Lambda response streaming — Lambda response streaming improves time to first byte and allows payloads up to 200 MB, versus 6 MB for buffered responses.
- Response streaming details — The first 6 MB of a streamed response is uncapped; after that the stream is limited to 2 MBps.
- 256 KiB payload limit — The input or output of a Step Functions task, state or execution is limited to 256 KiB.
- Standard vs Express — Standard workflows can run up to 1 year and keep up to 25,000 history events; Express workflows run at most 5 minutes.
- Human approval with a callback token — For human review, use a Standard workflow with .waitForTaskToken: the task pauses until the token returns through SendTaskSuccess or SendTaskFailure.
- Heartbeat on waiting tasks — A task waiting for a task token can wait until the one-year execution limit.
- Standard retry mode — AWS SDK standard retry mode (the default) retries transient and throttling errors with exponential backoff and full jitter, three attempts by default, and a retry quota that makes it fail fast when failures are widespread.
- Adaptive retry mode — Adaptive mode adds a client-side rate limiter that slows requests when throttling is detected, and it can delay even the initial request.
- Safeguard agent workflows — Task 2.1 expects safeguarded agent workflows: Step Functions stopping conditions, Lambda timeouts, IAM policies that enforce resource boundaries and circuit breakers.
- MCP servers on AWS — The exam guide places lightweight, stateless MCP servers on AWS Lambda and MCP servers that provide complex tools on Amazon ECS, with MCP client libraries to give consistent access patterns.
Guides that go deeper
- AIP-C01 Study Plan: Pass AWS GenAI Developer Pro
- Bedrock Provisioned Throughput vs Cross-Region Inference
Practice Domain 2
Answer practice questions, read the explanation and review your mistakes with spaced repetition.