Browse AWS Certification Guides

AIF-C01 Fundamentals of GenAI Guide

Study AIF-C01 Fundamentals of GenAI: key concepts, common traps, and exam decision cues.

This chapter covers the GenAI-specific part of the exam. AWS wants you to know what tokens, embeddings, multimodal models, and foundation model lifecycles mean, but it also wants business judgment about where GenAI helps and where its limitations change the answer.

Current weight in the exam guide

AWS currently weights Fundamentals of GenAI at 24% of scored content.

What this domain is really testing

This domain is testing whether you can talk about GenAI clearly without pretending every AI problem is a foundation-model problem. Strong answers here:

  • understand core GenAI vocabulary
  • know where GenAI creates business value and where it introduces risk
  • recognize the main AWS-managed GenAI service lanes without drifting into infrastructure detail

Work this domain in order

Lesson Focus
2.1 GenAI Concepts, Tokens, Embeddings & Models Learn the basic generative AI vocabulary and model patterns that show up throughout the exam.
2.2 GenAI Business Value, Limits & Model Selection Learn where GenAI creates value, where hallucinations or nondeterminism weaken it, and how to choose the right model type.
2.3 AWS GenAI Services, Infrastructure & Cost Trade-Offs Learn the AWS-managed GenAI toolset and the infrastructure or pricing trade-offs around it.

Fast routing inside this chapter

If the question is really about… Go first to…
tokens, embeddings, context windows, multimodal models, or what makes GenAI different 2.1 GenAI Concepts, Tokens, Embeddings & Models
whether GenAI is a good fit, what its limits are, or what kind of model to choose 2.2 GenAI Business Value, Limits & Model Selection
AWS Bedrock, managed services, cost, provisioning, or service fit 2.3 AWS GenAI Services, Infrastructure & Cost Trade-Offs

If you keep missing questions in this domain

Symptom What is usually going wrong Fix first
tokens, embeddings, and models blur together you are learning isolated terms instead of the role each plays in a GenAI workflow rework 2.1 and map concept to purpose
you keep choosing GenAI for the wrong use case you are ignoring hallucination, determinism, or cost constraints rework 2.2 and ask whether GenAI is actually the right business fit
AWS service questions feel like memorization you are not sorting services by job to be done rework 2.3 and classify each service as model access, app layer, enterprise assistant, or ML platform support
every answer seems innovative you are overvaluing novelty over fit and control prefer the answer that solves the stated problem with the least unnecessary complexity

What strong answers usually do

  • separate GenAI concept questions from AWS service-fit questions
  • recognize that GenAI outputs are probabilistic, not guaranteed deterministic truth
  • choose managed AWS paths when the question is really about business use rather than custom model engineering
  • weigh capability, latency, cost, governance, and hallucination risk together

Common AIF-C01 traps in this domain

  • treating embeddings like generated final answers instead of a representation used for search or similarity
  • assuming every chatbot question implies an agent
  • choosing a powerful model when the real need is lower cost, lower latency, or stricter control
  • confusing traditional predictive ML with text generation or conversational systems

Before you leave this domain

Make sure you can explain:

  1. what GenAI is doing in the scenario
  2. whether it is a good fit
  3. what major risk or limitation matters most
  4. which AWS-managed path fits the need best

Then move to 3. Applications of Foundation Models, where the exam starts asking how these concepts become real user-facing solutions.

In this section

Revised on Sunday, May 10, 2026
1. AI/ML Basics
3. FM Apps