How to Pass the AWS Certified AI Practitioner (AIF-C01) Exam in 2026: A 4-Week Study Plan

The AWS Certified AI Practitioner is the fastest-growing foundational certification in the AWS catalogue, and it is also the one most people underestimate. It carries the word "practitioner" and sits at foundational level, so candidates assume it is a gentler version of Cloud Practitioner with a few chatbot questions bolted on. It is not. Over half of the scored content is generative AI and foundation models, and that material is genuinely new for most people sitting it.

This guide gives you a 4-week study plan for the AWS Certified AI Practitioner (AIF-C01) exam, built around the actual domain weightings from the official exam guide rather than a generic reading list. If you can commit around eight hours a week, four weeks is enough to pass comfortably.

AIF-C01 Exam Format at a Glance

Before you plan anything, know exactly what you are walking into. These figures come from the current AWS exam guide.

Three things in that table change how you should study.

First, only 50 of the 65 questions count. The other 15 are unscored trial items that AWS is evaluating for future exams, and they are not flagged. If a question feels bizarre or oddly specific, it may well be one of them. Do not let it rattle you.

Second, the scoring model is compensatory. You do not need to pass each domain individually, only the exam overall. That means a genuinely weak area can be carried by strength elsewhere, so there is no need to reach perfection in every domain before you book.

Third, 90 minutes for 65 questions gives you roughly 83 seconds per question. That is comfortable for a foundational exam, but only if you are not translating unfamiliar terminology in your head on every item.

Exam Tip: The AIF-C01 uses four question formats: multiple choice, multiple response, ordering and matching. Ordering and matching are all-or-nothing. On a matching question with five pairs, four correct pairs earns you exactly the same as zero.

The Domain Weightings That Should Drive Your Study Plan

Most study plans give equal time to each domain. That is the single biggest mistake candidates make with this exam, because the AIF-C01 weightings are far from even.

Read that again. Domains 2 and 3 together are 52 percent of the scored content, roughly 26 of your 50 scored questions. Domain 3 alone is the largest single domain on the exam.

This is where candidates who come from a traditional cloud background get caught out. They arrive strong on IAM, S3 and the shared responsibility model, then discover that the exam mostly wants to know whether they understand prompt engineering, retrieval augmented generation, model evaluation metrics and when to fine-tune rather than prompt.

Your study time should follow the weightings. The plan below allocates roughly half of your effort to Domains 2 and 3.

Who This Exam Is Actually For

The official target candidate has up to six months of exposure to AI and ML technologies on AWS, and uses but does not necessarily build AI/ML solutions. That framing matters, because it tells you exactly how deep the questions go.

AWS explicitly puts the following outside the scope of the exam:

• Developing or coding AI/ML models and algorithms
• Data engineering and feature engineering techniques
• Hyperparameter tuning and model optimisation
• Building and deploying AI/ML pipelines or infrastructure
• Mathematical or statistical analysis of models
• Implementing security or compliance protocols for AI/ML systems
• Developing governance frameworks and policies

If you find yourself learning gradient descent maths or writing SageMaker training scripts, you have gone too deep and you are burning time you do not have. The exam wants selection and judgement, not implementation. It asks which service fits a business problem, not how to build it.

You are, however, expected to be familiar with core AWS services including Amazon EC2, Amazon S3, AWS Lambda, Amazon Bedrock and Amazon SageMaker AI, plus the shared responsibility model, IAM and AWS pricing models.

The 4-Week AIF-C01 Study Plan

This plan assumes around eight hours a week, which is two hours on four evenings, or a couple of longer weekend sessions. Total commitment is roughly 30 to 35 hours, which matches what most people who pass comfortably report.

Week 1: AI and ML Foundations Plus Responsible AI (Domains 1 and 4)

Start with the conceptual base, because everything in Domains 2 and 3 sits on top of it.

Days 1 and 2: core AI and ML concepts (Domain 1)

• The relationship between AI, machine learning, deep learning and generative AI
• Supervised, unsupervised and reinforcement learning, and which business problems suit each
• Inference versus training, and batch versus real-time inference
• Classification, regression, clustering and their typical use cases
• Model performance vocabulary: accuracy, precision, recall, F1, overfitting and underfitting
• Where AI is the wrong tool, which is a recurring question theme

Day 3: the AWS AI service map (Domain 1)

Learn what each managed service does in one sentence, and what distinguishes it from its neighbours. Amazon Rekognition for images and video, Amazon Transcribe for speech to text, Amazon Polly for text to speech, Amazon Comprehend for natural language processing, Amazon Translate, Amazon Textract for document extraction, Amazon Lex for conversational interfaces, Amazon Personalize for recommendations, Amazon Forecast for time series, Amazon Kendra for intelligent search, plus Amazon Bedrock, Amazon Q and Amazon SageMaker AI.

A large share of Domain 1 questions are simply scenario-to-service matching. Getting these automatic is the cheapest marks on the whole exam.

Day 4: Responsible AI (Domain 4)

Cover bias and fairness, transparency and explainability, robustness, veracity, and the difference between model bias and data bias. Learn what Amazon SageMaker Clarify does for bias detection and feature attribution, what model cards are for, and how AWS AI Service Cards document intended use and limitations. Domain 4 is only 14 percent, but it is high-yield because the concepts are small in number and phrased consistently.

Week 2: Generative AI Fundamentals (Domain 2)

This is 24 percent of the exam and the point where the material stops resembling other AWS foundational certs.

Days 1 and 2: generative AI concepts

• Tokens, embeddings, vectors, context windows and what a transformer architecture does at a conceptual level
• Foundation models versus large language models versus multimodal models
• Diffusion models for image generation
• Temperature, top-k and top-p, and what each one actually changes about output
• Hallucination: why it happens and how it is mitigated
• The generative AI lifecycle from model selection through evaluation to deployment

Day 3: Amazon Bedrock in depth

Bedrock is the centre of gravity for this exam. Know that it provides access to foundation models from multiple providers through a single API, and understand Bedrock Knowledge Bases, Bedrock Agents, Bedrock Guardrails, model evaluation and provisioned versus on-demand throughput. Understand where Amazon Q Business and Amazon Q Developer fit alongside it.

Day 4: benefits, limitations and cost

Learn the trade-offs: speed to market and lower barrier to entry against non-determinism, hallucination risk and inference cost. Understand the cost levers, including model choice, token volume, context length and provisioned throughput, because pricing judgement questions appear more often than candidates expect.

Week 3: Applications of Foundation Models (Domain 3)

The largest domain at 28 percent. Give it the most time, and be honest about which parts you cannot yet explain out loud.

Days 1 and 2: prompt engineering and RAG

• Zero-shot, few-shot and chain-of-thought prompting, and when each is appropriate
• Prompt templates, system prompts and negative prompting
• Prompt injection, jailbreaking and poisoning, plus the mitigations for each
• Retrieval augmented generation end to end: chunking, embedding, vector storage, retrieval, augmentation and generation
• The AWS vector store options, including Amazon OpenSearch Service, Amazon Aurora with pgvector and Amazon Neptune Analytics

RAG is the highest-value single topic on the exam. If you can explain when RAG solves a problem that fine-tuning cannot, and vice versa, you have covered a meaningful chunk of Domain 3.

Day 3: customisation and the decision tree

Understand the escalating ladder of model customisation and the cost, effort and data requirement at each step: prompt engineering first, then RAG, then fine-tuning, then continued pre-training, and training from scratch only as a last resort. Exam questions frequently describe a business scenario and ask for the most cost-effective approach. The answer is almost always the lowest rung of that ladder that satisfies the requirement.

Day 4: evaluation

Learn how model performance is judged: human evaluation, benchmark datasets, and the automated metrics ROUGE for summarisation, BLEU for translation and BERTScore for semantic similarity. Understand the business metrics too, including user satisfaction, cross-domain performance, conversion rate and average revenue per user.

Exam Tip: When a question asks for the "most cost-effective" or "least operational overhead" option, the correct answer is rarely fine-tuning and almost never training a model from scratch. Prompt engineering and RAG are the intended answers far more often.

Week 4: Security and Governance, Then Practice and Review (Domain 5)

Days 1 and 2: Domain 5

Cover IAM as it applies to AI workloads, encryption at rest and in transit, Amazon Macie for sensitive data discovery, VPC endpoints and PrivateLink for keeping inference traffic off the public internet, and the shared responsibility model applied to managed AI services. On governance, know data lineage, data cataloguing, model monitoring, AWS CloudTrail for auditing, AWS Config, and the major regulatory and standards touchpoints such as ISO, SOC and the EU AI Act as concepts rather than clause-level detail.

Days 3 and 4: full-length practice and targeted review

Sit full-length timed practice exams under real conditions, 65 questions in 90 minutes with no pauses. Then do the part that actually moves your score: review every question you got wrong and every question you got right by guessing, and write one sentence explaining why the correct answer is correct and why your choice was not.

Aim for a consistent 80 percent or better on practice exams before booking. That gives you enough headroom above the 700 scaled pass mark to absorb a bad question set on the day. You can drill AIF-C01 questions by domain and track your weak areas on the AWS Certified AI Practitioner practice exam on CertCrush.

Common Reasons People Fail AIF-C01

Having a plan is only half of it. These are the patterns that sink otherwise well-prepared candidates.

1. Studying Domain 1 hardest because it feels most familiar. It is only 20 percent. Domain 3 is 28 percent and gets a fraction of most people's attention.
2. Learning services as a list of names. The exam gives you a business scenario and expects you to pick a service. Names without use cases do not survive that translation.
3. Going too deep on ML theory. Coding, tuning and pipeline building are explicitly out of scope. Time spent there is time not spent on Bedrock and RAG.
4. Treating ordering and matching questions as partial credit. They are all-or-nothing, so an unfinished match is a zero.
5. Skipping Responsible AI as "soft" content. Domains 4 and 5 are 28 percent combined, exactly the same weight as Domain 3, and the concepts are far quicker to learn.
6. Booking too early after a single 70 percent practice attempt. The gap between 70 percent on practice and 700 scaled on the real thing is uncomfortably thin.

Exam Day Tactics

Do a first pass answering everything you know within about 60 seconds per question, and flag anything that needs thought. With 65 questions in 90 minutes, a disciplined first pass typically leaves you 20 minutes or more for the flagged items.

On multiple response questions, read the stem to find out how many answers are required before you read the options. On matching and ordering, do the pairs you are certain about first, then reason about the remainder by elimination.

Where two options both look defensible, pick the one that is more managed, lower cost and lower operational overhead. That is AWS's house style for foundational exams, and it resolves a surprising number of coin flips.

If you fail, AWS requires a 14-day wait before you can retake, and you pay again. That is a strong argument for not booking until your practice scores are consistently comfortable.

Frequently Asked Questions

How much does the AWS Certified AI Practitioner exam cost?

The AIF-C01 exam costs 100 USD. AWS periodically issues 50 percent retake or discount vouchers to people who have recently passed another AWS certification, so check your AWS Certification account before paying full price.

How hard is the AWS AI Practitioner exam?

It is foundational level, but it is harder than AWS Cloud Practitioner. The reason is content rather than depth: 52 percent of the scored questions cover generative AI and foundation models, which is unfamiliar territory for most candidates regardless of how much AWS experience they have. With around 30 to 35 hours of focused study it is very passable.

How long does it take to study for AIF-C01?

Most people who pass comfortably invest 25 to 35 hours. Four weeks at roughly eight hours a week fits that range. If you already work with generative AI day to day you can compress it to two or three weeks by front-loading Domains 1, 4 and 5.

Is the AWS AI Practitioner certification worth it?

For anyone in a cloud, data or business role who needs credible AI vocabulary, yes, and it is cheap and quick relative to associate-level certs. We covered the salary and career case in detail in Is the AWS AI Practitioner Certification Worth It in 2026?.

Does the AWS AI Practitioner certification expire?

Yes. The certification is valid for three years. You can recertify by retaking the exam, or it renews automatically if you pass a higher-level AWS certification before it lapses.

Should I take AWS AI Practitioner or Azure AI Fundamentals first?

It depends on your employer's cloud platform more than on the exams themselves. We compared them directly in AWS AI Practitioner vs Azure AI Fundamentals: Which First?.

Where AIF-C01 Fits in Your Certification Path

AIF-C01 has no prerequisites, so you can sit it as your first AWS certification. Most people pair it with AWS Certified Cloud Practitioner, in either order, to cover both general cloud literacy and AI literacy.

From there the natural next steps depend on your direction. If you want to build rather than advise, the AWS Certified Machine Learning Engineer Associate is the logical follow-on, though note that AWS is refreshing that exam, with beta registration for the updated version opening in September 2026. If you want to build generative AI applications specifically, look at the AWS Certified Generative AI Developer (AIP-C01). If your route is general cloud architecture, AWS Certified Solutions Architect Associate is the standard next move.

For a wider view of which AI credentials employers actually respond to, see AI Certifications Are Exploding, But Which Ones Actually Get You Hired?.

Ready to Start Practising?

Reading about Bedrock, RAG and prompt engineering is not the same as answering exam questions about them under time pressure. The candidates who pass AIF-C01 first time are the ones who spent their final week on realistic practice questions and reviewed every mistake properly.

CertCrush has an AWS Certified AI Practitioner (AIF-C01) question bank mapped to all five official domains, with full explanations for every answer so you learn from the questions you get wrong. Track your accuracy by domain, find the gaps, and close them before you book.

Create your free CertCrush account and start practising for AIF-C01 today.