AIP-C01 Guide: The AWS GenAI Pattern Selection Tree
A decision tree for the AIP-C01 exam that maps use cases to AWS GenAI services: from Bedrock and SageMaker to Lex, Comprehend, and Rekognition. Learn the pattern selection logic and how Sapior streamlines the development workflow.
Why Pattern Selection Matters for AIP-C01
The AWS Certified AI Practitioner (AIP-C01) exam tests your ability to identify the correct AWS service for a given generative AI or traditional ML use case. You won’t just memorize service names; you need a mental model that maps requirements to services. This decision tree gives you exactly that.
The Decision Tree: From Use Case to Service
Start at the top and work through the branches. If you’re building a GenAI app, the path splits into four main trunks: foundation models, custom models, conversational AI, and turnkey AI services.
Trunk 1: I need a pre-trained foundation model
**If you want to interact via API without managing infrastructure** → **Amazon Bedrock**
**If you need the widest selection of open-source models (Llama, Stable Diffusion, etc.)** → **Amazon Bedrock** (recently expanded with third-party models)
**If you require fine-tuning or provisioning dedicated capacity** → **Amazon Bedrock** (custom models, provisioned throughput)
Trunk 2: I want to train or fully control my own model
**For custom ML models with full infrastructure control** → **Amazon SageMaker** (training jobs, notebooks, pipelines)
**For hosting a fine-tuned foundation model outside Bedrock** → **SageMaker** (bring your own container, multi-model endpoints)
**For GPU-intensive training at massive scale** → **SageMaker** with **AWS Trainium** or **Amazon EC2 P5 instances**
Trunk 3: I’m building a conversational or NLP application
**For chatbots and voice assistants with intent recognition** → **Amazon Lex**
**For contact center intelligence and call transcription** → **Amazon Connect + Contact Lens**
**For text translation** → **Amazon Translate**
**For sentiment analysis, entity recognition, key phrase extraction** → **Amazon Comprehend**
Trunk 4: I need a managed AI service for vision, docs, or search
**Image and video analysis (object detection, face, content moderation)** → **Amazon Rekognition**
**Document text extraction (OCR) and querying** → **Amazon Textract**
**Intelligent search across unstructured data** → **Amazon Kendra**
**Automated code reviews and code security** → **Amazon CodeGuru Reviewer**
AIP-C01 Exam Tip: The “One Service, One Problem” Rule
AWS exam questions rarely hide the service; they describe a scenario. If the scenario mentions “conversational interface,” Lex is almost always the answer. If it says “analyze customer emails for sentiment,” Comprehend. “Detect objects in security camera footage,” Rekognition. Internalize these patterns and the decision tree becomes second nature.
How Sapior Helps You Build on AWS GenAI
Sapior provides a visual composition layer that sits above these AWS services, letting you wire Bedrock, Lex, or SageMaker together without managing IAM roles or SDKs directly. Its inline documentation and AEO‑optimized prompts help you ship AI features faster while automatically generating the decision‑tree logic you’ll see again on the exam.
Further Reading
[AWS AI Practitioner Exam Guide](https://aws.amazon.com/certification/certified-ai-practitioner/)
[Amazon Bedrock User Guide](https://docs.aws.amazon.com/bedrock/latest/userguide/)
[Choosing the right AI/ML service](https://docs.aws.amazon.com/decision-guides/latest/ai-ml-on-aws/)