Certification guide
AIF-C01AWS Certified AI Practitioner: the honest guide
AWS Certified AI Practitioner is the AI counterpart to Cloud Practitioner, and it behaves like one. It asks what a foundation model is, when Amazon Bedrock is the right answer to a described problem, what retrieval augmented generation does to a model's output, and who is accountable when that output is wrong. It never asks you to build or tune anything.
The format is 65 questions in 90 minutes. Fifty of those are scored and 15 are unscored items AWS is trialling for future versions of the exam. Nothing on screen marks the difference, so you answer all 65 as though they count, and your pacing budget is 65 questions rather than 50: a bit over 80 seconds each. People who hear 'only 50 count' and slow down are the ones who run out of clock.
The pass mark is 700 on a scale of 100 to 1000. That is a scaled score, not a percentage, so 700 does not mean 70 percent of the questions right. AWS sets the scale so a score means the same thing across different versions of the exam, which means the raw number of correct answers you need moves with the form you sit.
Alongside multiple choice and multiple response, AIF-C01 uses the newer AWS item types: ordering, matching, and case studies where several questions hang off one scenario. None of them are difficult once you have seen one. All of them are unwelcome the first time, and exam day is a bad first time.
Who AI Practitioner is for
A good fit if
- You work near an AI or generative AI project without building it: product, program, support, marketing or sales.
- You already hold Cloud Practitioner or an AWS associate certification and want the AI vocabulary that goes with it.
- You sit in the review path for AI work in security, risk, procurement, legal or compliance, and you need to ask better questions than a vendor deck answers.
- You are heading toward ML Engineer Associate and want the foundation AWS names as the step before it.
- Your organization has decided to use generative AI and nobody in the room can reliably tell a sound Bedrock use case from a bad one.
Probably not, if
- You want to build, fine-tune and deploy models as your actual job. AIF-C01 stops at the concept and the service name. AWS Certified Machine Learning Engineer Associate is the engineering exam, and this one is the recommended step before it rather than a smaller substitute for it.
- You have never used AWS. This exam assumes cloud vocabulary it does not teach, and the AWS page for it points people new to IT and AWS at Cloud Practitioner material first. Start there, even if you never sit CLF-C02.
- You already work as a data scientist or an ML engineer. You will pass this over a weekend and learn close to nothing, and the credential says less about you than the work on your CV already does.
- You want a coding exam. There are no notebooks, no code to read and no configuration to write. Every question is a concept or a service-selection decision.
Is AI Practitioner worth it?
For someone working near AI without building it, yes. The specific gap AIF-C01 closes is between being able to say 'generative AI' and being able to say why a foundation model is the wrong tool for this particular problem. That gap costs organizations real money in projects that should never have been started, and $100 with a few weeks of evenings is a cheap correction for it.
For an engineer heading to ML Engineer Associate, the case is structural and financial rather than educational. You will not learn much, but AWS names AI Practitioner as the foundation before MLA-C01, and passing it earns a 50 percent voucher toward that next exam, which recovers half the fee straight away. If you have no interest in the ML track, that argument disappears and so does most of the reason to sit it.
For someone hoping it produces an AI job on its own, no, and it would be dishonest to suggest otherwise. It is a foundational certification and hiring managers read it as exactly that. What it does is stop a screening filter that says 'AI familiarity' from rejecting you, and give you standing in a conversation you were previously listening to. The role change comes from the work, or from the associate exam after this one.
What the exam actually asks you to do
Multiple choice and multiple response, plus the newer AWS item types: ordering, matching, and case studies with several questions on one scenario.
Item formats
- Multiple choice
- Multiple response
- Drag and drop
The highlighted formats are the ones you cannot answer from memory alone. AWS, AI Practitioner exam guide ↗
Domain breakdown and official weightings
From the official AWS exam guides. Applications of Foundation Models is the heaviest domain at 28 percent, followed by Fundamentals of Generative AI at 24 percent.
- Fundamentals of AI and ML20%
- Fundamentals of Generative AI24%
- Applications of Foundation Models28%
- Guidelines for Responsible AI14%
- Security, Compliance, and Governance for AI Solutions14%
Study plans by experience level
No AWS and no AI background
4 to 5 weeksat 5 to 6 hours
- 1Week 1: cloud and AI vocabulary before anything else. What a model is, what training and inference are, and what AWS's basic building blocks are called. AIF-C01 assumes all of it and teaches none of it, so this week is the one that decides how hard the rest feels.
- 2Week 2: Fundamentals of AI and ML, 20 percent of the exam and the base the other four domains stand on. Supervised against unsupervised, training against inference, and the cases where machine learning is the wrong tool entirely.
- 3Week 3: Fundamentals of Generative AI, 24 percent. Foundation models, tokens, embeddings, prompt engineering, and a clear-eyed account of what the technology cannot do. Hallucination and non-determinism are examined, not glossed over.
- 4Week 4: Applications of Foundation Models, 28 percent and the heaviest domain on the exam. Amazon Bedrock, choosing a model for a described need, retrieval augmented generation as a pattern, and when fine-tuning earns its cost.
- 5Week 5: Guidelines for Responsible AI plus Security, Compliance, and Governance for AI Solutions, 14 percent each and 28 percent together. Then AWS's free sample questions until the phrasing is unremarkable, and book.
Holds Cloud Practitioner or an AWS associate certification
1 to 2 weeksat 6 to 8 hours
- 1Skim the AI and ML fundamentals domain rather than studying it. Your AWS grounding already covers the shared responsibility habits and the service-selection reasoning this exam rewards.
- 2Spend the first block on the two generative AI domains, 52 percent of the exam between them and the part your existing certification did not touch.
- 3Learn Bedrock properly: what it is for, how model choice gets framed, and where retrieval augmented generation sits against fine-tuning for a described requirement. The service-selection questions live here.
- 4Give one full session to responsible AI and one to AI security and governance. They are 14 percent each, they are examined precisely rather than woolly, and they are the most under-studied part of the exam because they read like policy.
- 5Work the free AWS sample questions, sit the official practice exam if your Skill Builder subscription includes it, and book once the wording stops catching you out.
Non-technical, working alongside an AI project
5 to 6 weeksat 4 hours
- 1Weeks 1 to 2: AI and ML fundamentals, slowly. This is the domain that changes your working conversations most and the one you can least afford to rush.
- 2Week 3: generative AI fundamentals, focused on the boundaries of the technology. Hallucination, non-determinism, and what a foundation model does not know are the parts you will use in a meeting the following Monday.
- 3Week 4: applications of foundation models. Read it as a catalog of when to reach for what rather than as a technical specification, because that framing is also how the questions are written.
- 4Week 5: responsible AI plus security, compliance and governance. This section overlaps most directly with a risk, legal or procurement role, so it earns its time twice over.
- 5Week 6: practice questions, paying attention to how AWS phrases a scenario rather than to the technology inside it. The phrasing is consistent and learnable, and it is most of what separates a pass from a near miss at this level.
Data or software engineer who needs the credential on paper
1 weekat 4 to 5 hours
- 1Read the exam guide and mark only the AWS service names and the governance vocabulary. That short list is your syllabus. The AI concepts you already have.
- 2Learn where AWS draws its own lines: which service it names for which need, and how it frames responsible AI. The exam wants AWS's answer, which is not always the answer you would give.
- 3Sit a full timed practice set early in the week rather than at the end. It will find the compliance and governance terminology you assumed you knew, which is where working engineers lose marks on this exam.
- 4Fix what that set exposed, then book. Resist re-reading the machine learning fundamentals domain: it is 20 percent of the exam and the part you are least likely to lose anything on.
Common mistakes
- Studying it like an engineering exam
- Anyone technical is tempted to go deep on architectures, training and tuning. AIF-C01 asks which service fits a described need and what a concept means. Depth on tuning is time not spent on the governance vocabulary that actually appears on the paper.
- Treating the two generative AI domains as one topic to skim
- Fundamentals of Generative AI is 24 percent and Applications of Foundation Models is 28 percent. That is 52 percent of the exam sitting in the two domains people assume they already know because they have used a chatbot.
- Dismissing responsible AI and governance as soft content
- The two domains are 14 percent each and 28 percent together, and they are examined with precision: bias and fairness, explainability, data handling, and who is accountable for a model's output. One read-through is not preparation.
- Reading 700 as 70 percent
- The pass mark is 700 on a 100 to 1000 scale. It is scaled, so it does not map to a fixed number of correct answers, and aiming at 70 percent sets a target that does not exist on this exam.
- Pacing to 50 questions instead of 65
- Fifty questions are scored and 15 are not, and nothing tells you which is which. You answer all 65, so the clock is 90 minutes across 65 items, a bit over 80 seconds each. Anyone budgeting for 50 finishes the paper in a hurry.
- Meeting the ordering and case study formats for the first time on exam day
- Multiple choice is not the whole exam. Ordering, matching, and case studies with several questions hanging off one scenario all appear. The case study format rewards reading the scenario once, carefully, rather than four times under pressure.
What comes after passing
AI Practitioner is valid for three years. Recertifying means passing the current version of the exam, or passing a higher certification in the same track. Nothing recurs between those points: no annual fee, no continuing education units, no maintenance requirement.
Check your AWS Certification account within a few days of passing. The benefits from a pass include a 50 percent discount voucher toward your next AWS exam, and it carries an expiry date. Sitting the next exam while that voucher is live is both the intended path and the cheapest one, and letting it lapse is the most common way people pay full price twice.
The step up from here is AWS Certified Machine Learning Engineer Associate, which AWS lists AI Practitioner as the recommended foundation for. It is a real change of level: MLA-C01 is an engineering exam about data preparation, model development, deployment and monitoring, and passing it recertifies this one at the same time. If you came to AIF-C01 without much AWS behind you, Cloud Practitioner is the sideways move worth making before the upward one, because it fills the general cloud grounding this exam assumed you had.
At work, the honest expectation is that this credential changes conversations rather than job titles. It gives you standing to ask why a project chose a foundation model over something simpler, and vocabulary that holds up when a vendor is presenting. That is genuinely useful and it is not a promotion.
Where people go next
Costs across the full renewal cycle are on the AI Practitioner cost page.
Frequently asked questions
Is AWS Certified AI Practitioner worth it?
For people working near AI without building it, yes: $100 buys the vocabulary to tell a sound generative AI use case from a bad one, which is the expensive mistake at this level. For engineers heading to ML Engineer Associate, the argument is mostly the 50 percent voucher toward that exam. For anyone expecting it to produce an AI job on its own, no.
How many questions are on the AIF-C01 exam?
65 questions in 90 minutes. Fifty of them are scored and 15 are unscored items AWS is trialling for future versions, and nothing identifies which is which, so you answer all 65 as though every one counts. Pace to 65, not to 50.
What score do I need to pass AIF-C01?
700 on a scale of 100 to 1000. It is a scaled score rather than a percentage, so 700 does not mean getting 70 percent of the questions right. The raw number of correct answers required shifts with the version of the exam you sit, which is the point of scaling it.
How long does AIF-C01 take to study for?
One to two weeks at 6 to 8 hours if you already hold Cloud Practitioner or an AWS associate certification. Four to five weeks from no AWS and no AI background. Five to six weeks at a lighter pace if you are non-technical and studying around a full working week.
Do I need Cloud Practitioner before AIF-C01?
No, there is no prerequisite, and the registry records none for this exam. AIF-C01 does assume basic AWS vocabulary though, and the AWS page for it points people new to IT and AWS at Cloud Practitioner material first. Working through that material makes this exam easier even if you never sit CLF-C02.
Does AWS Certified AI Practitioner expire?
Yes. AWS certifications last three years. You recertify by passing the current version of the exam again, or by passing a higher certification in the same track, and there is no annual fee or continuing education requirement in between. Passing ML Engineer Associate recertifies this one.
How hard is AIF-C01?
It is a foundational exam and a fair one. The difficulty is breadth of vocabulary rather than technical depth, and the marks most often go missing in the responsible AI and the security and governance domains, which read like policy and are examined more precisely than people expect.
Keep reading
- AI Practitioner practice questions
Free sample questions with the full explanation on every answer.
- Free AI Practitioner practice test
10 original questions, playable now. No account, no card.
- AI Practitioner exam format
How many questions, how long, what the items look like, and how long it stays valid.
- AI Practitioner passing score
The exact cut score, what kind of number it is, and the retake terms.
- How hard is AI Practitioner?
An honest difficulty read from the format, the clock and the weights.
- What AI Practitioner costs
The voucher price, the retake, and what renewal costs across the cycle.
Practice AI Practitioner for free while you decide
Original questions written from the published objectives, with the concept, the reasoning, and a note on every wrong option. No account needed to start.
Start free AI Practitioner questions