Implementation and integration cheat sheet
All 16 AIP-C01 flashcards for Implementation and integration on one page, grouped by exam objective. Print it or save it as a PDF for a last-minute read.
Implement agentic AI solutions and tool integrations
- What is an action group in Bedrock Agents?
- A set of actions an agent can take, defined with an OpenAPI schema or function details and backed by a Lambda function (or returned to your app). Amazon Bedrock Agents Classic is closed to new customers; new builds use Amazon Bedrock AgentCore.
- What does the Model Context Protocol standardise?
- How AI clients discover and call tools and data sources, so one tool server works with many agents.
- Lambda or ECS for an MCP server?
- Lambda for stateless, short, lightweight tools. ECS for complex, long-running or stateful ones.
- What is the ReAct pattern?
- Alternate reasoning steps with actions (tool calls), feeding each observation back into the next reasoning step.
- Name four safeguards for an agent loop.
- Max iterations, timeouts, least-privilege IAM, and a circuit breaker. Add human approval for risky actions.
Implement model deployment strategies
- What does Provisioned Throughput buy you?
- Reserved model units for predictable throughput and latency, billed whether or not you use them.
- What is model cascading?
- Send routine queries to a smaller, cheaper model and escalate only the hard ones to a bigger model.
- When choose SageMaker AI endpoints over Bedrock?
- When you need a model Bedrock does not offer, deep control of the container and hardware, or a self-hosted fine-tuned model.
Design and implement enterprise integration architectures
- Which event service decouples a legacy app from GenAI features?
- Amazon EventBridge, optionally with SQS for buffering slow model jobs.
- How do you grant a team access to some models but not all?
- Use IAM policies that allow bedrock:InvokeModel only on specific model or inference profile ARNs.
- What belongs in a GenAI CI/CD pipeline?
- Prompt and tool tests, security scans, evaluation quality gates, and rollback support, run by CodePipeline and CodeBuild.
Implement FM API integrations
- Why put API Gateway in front of a model call?
- Request validation, throttling and usage plans, authorisation, and a stable contract for clients.
- How should you handle transient model errors?
- Retry with exponential backoff and jitter, cap attempts, then fall back or degrade gracefully.
- What is static versus dynamic model routing?
- Static routing is fixed in code or config. Dynamic routing picks a model per request from content or live metrics, for example with Step Functions or intelligent prompt routing.
Implement application integration patterns and development tools
- What does Bedrock Prompt Flows give non-developers?
- A visual, low-code builder for chaining prompts, conditions, knowledge bases and Lambda steps into a workflow.
- Where do Logs Insights and X-Ray fit when debugging an FM app?
- Logs Insights queries invocation logs for prompts and responses, and X-Ray traces the call path across services and the model API.