Official exam guide summary - AIP-C01
Paraphrased from AWS's own pages. Task-statement titles are quoted as AWS wrote them; the skill notes under each are summaries, not copies. Read the guide itself for the full skill wording.
Sources (all fetched 2026-10-30, HTTP 200)
Exam mechanics
| Fact | Value | Source |
|---|---|---|
| Total questions | 75 | C (format line); G (65 scored + 10 unscored) |
| Scored / unscored | 65 scored, 10 unscored; unscored are not identified | G |
| Question types | Multiple choice (1 correct of 4) and multiple response (2+ correct of 5+, all must be selected) | G; C |
| Guessing | Unanswered = wrong; no penalty for guessing | G |
| Duration | 180 minutes | C |
| Passing score | 750 on a 100-1,000 scaled score; pass/fail; compensatory (no per-section pass needed) | G |
| Cost | 300 USD per attempt (Professional tier) | C; P |
| Delivery | Pearson VUE test center or online proctored | C |
| Languages | English, Japanese, Korean, Simplified Chinese | C |
| Retake | 14-day wait after a fail; full fee each attempt | F |
| Validity | Recertify every 3 years | F |
| Prerequisite certs | None required; AWS lists AI Practitioner, Solutions Architect - Associate, ML Engineer - Associate, Data Engineer - Associate as helpful | C |
| Beta / retirement status | No beta or retirement notice on G or C as of 2026-10-30 | G; C |
Target candidate (G, C): 2+ years building production apps on AWS or with open-source tech, general AI/ML or data engineering experience, and 1 year hands-on GenAI. Recommended AWS knowledge: compute/storage/networking, security and identity, deployment and IaC, monitoring and observability, cost optimization.
Out-of-scope job tasks (G): model development and training, advanced ML techniques, data engineering and feature engineering.
Section feedback (G): the score report may show per-section classifications; AWS warns to interpret them with caution.
Tag key
The founder has not stated his AWS hands-on depth. Assumption: GenAI concepts are strong (3 AI-adjacent certs passed, daily Claude agent/harness work, Snowflake Cortex), and AWS service mechanics are the likely gap. The pretest tests this assumption (see study-plan.md).
[likely-known]- a concept he already exercises in Claude or Snowflake work; AWS adds a thin naming layer.[AWS-gap]- the questions turn on AWS service selection, configuration or API behavior.[mixed]- known concept, but the skill examples lean on AWS-specific mechanics.
Tag count: 4 likely-known, 10 mixed, 6 AWS-gap.
Domain 1: Foundation Model Integration, Data Management, and Compliance - 31% (G, D1)
| Task | Title (AWS wording) | Tag | Why |
|---|---|---|---|
| 1.1 | Analyze requirements and design GenAI solutions. | [mixed] |
Solution design is his daily work; the AWS layer is Bedrock proof-of-concept work and the Well-Architected Generative AI Lens. |
| 1.2 | Select and configure FMs. | [AWS-gap] |
Model choice is known, but the skills are AppConfig-driven model switching, Step Functions circuit breakers, Bedrock cross-Region inference, and SageMaker AI LoRA deployment plus Model Registry. |
| 1.3 | Implement data validation and processing pipelines for FM consumption. | [mixed] |
Data quality pipelines are his strength; Glue Data Quality, Data Wrangler, Transcribe, Comprehend and Bedrock request JSON formats are AWS-specific. |
| 1.4 | Design and implement vector store solutions. | [mixed] |
Vector store design is known from Cortex Search and RAG work; OpenSearch Neural plugin, sharding, Aurora pgvector and DynamoDB metadata patterns are AWS-specific. |
| 1.5 | Design retrieval mechanisms for FM augmentation. | [likely-known] |
Chunking, embedding choice, hybrid search, reranking, query decomposition and MCP retrieval are core RAG skills he uses; learn Bedrock Knowledge Bases chunking option names (beta-exam colleague flagged chunking). |
| 1.6 | Implement prompt engineering strategies and governance for FM interactions. | [mixed] |
Prompt design, chain-of-thought, output specs and prompt regression testing are known; Bedrock Prompt Management versions and approval, Prompt Flows and CloudTrail tracking are AWS mechanics. |
Skill count: 1.1 (3), 1.2 (4), 1.3 (4), 1.4 (5), 1.5 (6), 1.6 (6).
Domain 2: Implementation and Integration - 26% (G, D2)
| Task | Title (AWS wording) | Tag | Why |
|---|---|---|---|
| 2.1 | Implement agentic AI solutions and tool integrations. | [mixed] |
Agents, MCP, tool validation, stopping conditions and human review are his home turf; Strands Agents, AWS Agent Squad, Step Functions ReAct and Lambda- or ECS-hosted MCP servers are AWS-specific. |
| 2.2 | Implement model deployment strategies. | [AWS-gap] |
Claude API and Cortex hide hosting; Bedrock provisioned throughput vs on-demand, SageMaker AI endpoints and GPU container loading are new surface. |
| 2.3 | Design and implement enterprise integration architectures. | [AWS-gap] |
Event-driven design is general, but answers hinge on EventBridge, API Gateway, identity federation, Outposts, Wavelength, CodePipeline and CodeBuild selection. |
| 2.4 | Implement FM API integrations. | [mixed] |
Streaming, retries with backoff and model routing are known from the Claude API; Bedrock streaming APIs, SQS async, API Gateway rate limiting and X-Ray are AWS mechanics. |
| 2.5 | Implement application integration patterns and development tools. | [AWS-gap] |
Tool-name heavy: Amplify, OpenAPI on API Gateway, Bedrock Data Automation, Amazon Q Developer, CloudWatch Logs Insights. |
Skill count: 2.1 (7), 2.2 (3), 2.3 (5), 2.4 (4), 2.5 (6).
Domain 3: AI Safety, Security, and Governance - 20% (G, D3)
| Task | Title (AWS wording) | Tag | Why |
|---|---|---|---|
| 3.1 | Implement input and output safety controls. | [mixed] |
Guardrails, injection defense and schema-enforced output are things he builds in harnesses; Bedrock Guardrails filter types and Comprehend pre-filters are AWS-specific (beta-exam colleague flagged Guardrails). |
| 3.2 | Implement data security and privacy controls. | [AWS-gap] |
VPC endpoints, IAM policies, Lake Formation, Macie and S3 Lifecycle are AWS security mechanics; IAM is on AWS's recommended-knowledge list. |
| 3.3 | Implement AI governance and compliance mechanisms. | [mixed] |
Governance principles are known (Snowflake GES-C02 had a governance domain); SageMaker AI model cards, Glue Data Catalog lineage and CloudTrail audit are AWS-specific. |
| 3.4 | Implement responsible AI principles. | [likely-known] |
Transparency, fairness evaluation, LLM-as-a-judge and model cards are concept-level; Bedrock agent tracing is the one AWS item to learn. |
Skill count: 3.1 (5), 3.2 (3), 3.3 (4), 3.4 (3).
Domain 4: Operational Efficiency and Optimization for GenAI Applications - 12% (G, D4)
| Task | Title (AWS wording) | Tag | Why |
|---|---|---|---|
| 4.1 | Implement cost optimization and resource efficiency strategies. | [likely-known] |
Token budgets, context pruning, tiered model use and prompt caching are context-engineering basics for him; provisioned throughput sizing is the AWS piece. |
| 4.2 | Optimize application performance. | [mixed] |
Latency tradeoffs, streaming and temperature/top-p tuning are known; latency-optimized Bedrock models, batch inference and auto-scaling are AWS-specific (beta-exam colleague flagged latency). |
| 4.3 | Implement monitoring systems for GenAI applications. | [AWS-gap] |
Golden datasets and tracing ideas are known, but the skills ask which AWS signal captures what: CloudWatch metrics and anomaly detection, Bedrock Model Invocation Logs, cost anomaly detection (colleague flagged observability). |
Skill count: 4.1 (4), 4.2 (6), 4.3 (6).
Domain 5: Testing, Validation, and Troubleshooting - 11% (G, D5)
| Task | Title (AWS wording) | Tag | Why |
|---|---|---|---|
| 5.1 | Implement evaluation systems for GenAI. | [likely-known] |
Eval design, RAG evaluation, LLM-as-a-judge, regression gates and agent task-completion metrics are his eval work; learn Bedrock Model Evaluations and Agent evaluations by name (colleague flagged model/agent evaluation). |
| 5.2 | Troubleshoot GenAI applications. | [mixed] |
Context overflow, truncation and retrieval drift diagnosis are known; the tools named are CloudWatch Logs, X-Ray and schema validation. |
Skill count: 5.1 (9), 5.2 (5).
Totals: 5 domains, 20 task statements, 98 skills (D1-D5).
In-scope AWS services (IS)
AWS calls the list non-exhaustive and subject to change. AWS gives no weight signal for services, and the technologies page (TC) says list order means nothing. Stars below are Ray-derived, from how often each service appears in the 98 skill statements (D1-D5), not an AWS signal. ★ = 5+ mentions, ☆ = 2-4 mentions.
- Machine Learning: ★ Amazon Bedrock (38 mentions across features), ★ Amazon Bedrock Guardrails (6, listed under Bedrock), ☆ Bedrock Knowledge Bases, ☆ Bedrock Prompt Management, ☆ Bedrock Prompt Flows, ★ Amazon SageMaker AI (7), ★ Amazon Comprehend (5), ☆ Amazon Q Developer, Amazon Bedrock AgentCore, Amazon Augmented AI, Amazon Kendra, Amazon Lex, Amazon Q Business, Amazon Q Business Apps, Amazon Quick, Amazon Rekognition, SageMaker Clarify, SageMaker Data Wrangler, SageMaker Ground Truth, SageMaker JumpStart, SageMaker Model Monitor, SageMaker Model Registry, SageMaker Neo, SageMaker Processing, SageMaker Unified Studio, Amazon Textract, Amazon Titan, Amazon Transcribe
- Compute: ★ AWS Lambda (16), AWS App Runner, Amazon EC2, Lambda@Edge, AWS Outposts, AWS Wavelength
- Application Integration: ★ AWS Step Functions (11), AWS AppConfig (1, but flagged by beta-exam colleague), Amazon AppFlow, Amazon EventBridge, Amazon SNS, Amazon SQS
- Management and Governance: ★ Amazon CloudWatch (10, incl. CloudWatch Logs 4), ☆ AWS CloudTrail, AWS Auto Scaling, AWS Chatbot, Amazon CloudWatch Synthetics, AWS Cost Anomaly Detection, AWS Cost Explorer, Amazon Managed Grafana, AWS Service Catalog, AWS Systems Manager, AWS Well-Architected Tool
- Networking and Content Delivery: ★ Amazon API Gateway (9), AWS AppSync, Amazon CloudFront, Elastic Load Balancing, AWS Global Accelerator, AWS PrivateLink, Amazon Route 53, Amazon VPC
- Analytics: ☆ Amazon OpenSearch Service (4), ☆ AWS Glue (3), Amazon Athena, Amazon EMR, Amazon Kinesis, Amazon Quick Sight, Amazon MSK
- Database: ☆ Amazon DynamoDB, Amazon Aurora, Amazon DocumentDB, DynamoDB Streams, Amazon ElastiCache, Amazon Neptune, Amazon RDS
- Developer Tools: ☆ AWS X-Ray (3), AWS Amplify, AWS CDK, AWS CLI, AWS CloudFormation, AWS CodeArtifact, AWS CodeBuild, AWS CodeDeploy, AWS CodePipeline, Kiro, AWS Tools and SDKs
- Security, Identity, and Compliance: ☆ IAM (2), Amazon Cognito, AWS Encryption SDK, IAM Access Analyzer, IAM Identity Center, AWS KMS, Amazon Macie, AWS Secrets Manager, AWS WAF
- Containers: Amazon ECR, Amazon ECS, Amazon EKS, AWS Fargate
- Storage: ☆ Amazon S3 (4), Amazon EBS, Amazon EFS, S3 Intelligent-Tiering, S3 Lifecycle policies, S3 Cross-Region Replication
- Migration and Transfer: AWS DataSync, AWS Transfer Family
- Customer Engagement: Amazon Connect
Named in skills but not on the in-scope list: Strands Agents and AWS Agent Squad (D2, skills 2.1.1, 2.1.6, 2.5.5), plus MCP (6 mentions). These are open-source frameworks or protocols rather than AWS services, so expect them as answer options.
On the in-scope list but never named in a skill: Amazon Bedrock AgentCore and Kiro, among others. Low expected weight, but they can appear as options.
Out-of-scope AWS services (OOS)
Non-exhaustive; AWS excludes services unrelated to the role entirely. Items most likely to show up as distractors for this candidate:
- Analytics: AWS Clean Rooms, AWS Data Exchange, Amazon DataZone, Amazon FinSpace
- Database: Amazon Redshift (out of scope, worth noting for a data engineer), Amazon Keyspaces, Amazon QLDB, Amazon Timestream
- Machine Learning: Amazon DevOps Guru, Amazon Forecast, Amazon Fraud Detector, Amazon HealthLake, Lookout for Equipment/Metrics/Vision, Amazon Monitron, AWS Panorama, AWS DeepComposer, AWS DeepRacer
- Compute: AWS Batch, EC2 Image Builder, ECS Anywhere, EKS Anywhere, Elastic Beanstalk, Lightsail, AWS Local Zones, Serverless Application Repository
- Cloud Financial Management: AWS Budgets, Cost and Usage Report, Reserved Instance reports, Savings Plans
- Networking: App Mesh, Cloud Map, Direct Connect, Private 5G, Transit Gateway, AWS VPN
- Developer Tools: Cloud9, CloudShell, CodeGuru, CodeStar, Corretto
- Management and Governance: Console Mobile App, Health Dashboard, License Manager, Proton, Trusted Advisor
- Application Integration: Amazon MQ
- Customer Engagement: Amazon SES
- Also out: the full IoT, Media Services (incl. Kinesis Video Streams), End User Computing, Frontend Web and Mobile (Device Farm, Location Service, Pinpoint), Game Development, Blockchain, Business Applications, Migration (Snow Family, Migration Hub, Application Migration Service), Quantum, Robotics and Satellite categories.
Naming notes
- The exam uses official short names (for example "Amazon SNS"); the in-exam Help feature lists short-to-full names (SN).
- The guide says "Amazon Bedrock Prompt Flows". Current Bedrock docs title that feature "Amazon Bedrock Flows" (https://docs.aws.amazon.com/bedrock/latest/userguide/flows.html). Treat them as the same thing.