CCA vs AWS AI Practitioner: Which Certification Should You Take?

The CCA-F and AWS Certified AI Practitioner answer different questions: the CCA-F certifies that you can make sound design decisions when building agents on Claude (loops, tools, MCP, Claude Code, structured output, reliability), while AWS AI Practitioner certifies broad AI and machine-learning literacy in the context of AWS services. Engineers building on Claude should take the CCA-F; people who need general AI vocabulary or work in an AWS-centric organisation should take the AWS exam; many will reasonably do both.

Here is a side-by-side, with a rule for choosing. CCA-F details are as of mid-2026; AWS details change and are stated only where they are public and stable, so confirm both on the official Anthropic / Pearson VUE page and AWS’s certification page before you register.

What does each certification actually cover?

Claude Certified Architect – Foundations (CCA-F, exam code CCAR-F). Issued by Anthropic, one of four exams in the Claude Certification Program. Five domains: D1 Agentic Architecture & Orchestration 27%, D2 Tool Design & MCP Integration 18%, D3 Claude Code Configuration & Workflows 20%, D4 Prompt Engineering & Structured Output 20%, D5 Context Management & Reliability 15%. Every question is a scenario asking which design is least bad under constraints. It is vendor-specific and product-specific: Claude API, Agent SDK, Claude Code and MCP.

AWS Certified AI Practitioner. Issued by AWS as a foundational-level certification. Its published domains cover fundamentals of AI and ML, fundamentals of generative AI, applications of foundation models, guidelines for responsible AI, and security, compliance and governance for AI solutions, all framed around AWS services. It is a literacy exam: it does not require you to have built an agent, and it is not specific to any one model vendor. Check AWS’s page for the current exam guide.

The one-line difference: CCA-F is depth on one stack’s architecture; AWS AI Practitioner is breadth on AI concepts and one cloud’s services.

How do the formats compare?

	CCA-F	AWS Certified AI Practitioner
Issuer	Anthropic	Amazon Web Services
Level	Foundations tier of the Architect track	Foundational (AWS’s entry tier)
Question style	60 scenario-based multiple-choice / multiple-response, 120 minutes, closed book	Multiple-choice and multiple-response plus some newer item types; check AWS’s page for count and duration
Passing	Scaled 720 on 100–1000	Scaled score; check AWS’s page
Fee	US$125	Foundational-tier fee set by AWS; check AWS’s page
Delivery	Pearson VUE, online proctored or test center	Pearson VUE, online or test center
Validity	12 months	AWS certifications are typically valid for three years; confirm on AWS’s page
Assumes hands-on building?	Yes	No

We deliberately do not state AWS numbers we cannot guarantee are current. The structural contrast is stable: CCA-F is shorter-lived, dearer per year, and much more specific.

Who should take the CCA-F?

• Engineers who build or will build on the Claude API, Agent SDK or Claude Code in the next year.
• Solutions architects and consultants selling Claude-based designs, who need a checkable vendor credential.
• Tech leads standardising Claude Code across a team; Domain 3 alone justifies the study.
• Anyone who wants a forcing function to learn agent architecture rather than only prompting.

Skip it if you use Claude only through the chat app, or your organisation is committed to another model vendor. See who should get the CCA-F.

Who should take AWS AI Practitioner?

• Product managers, analysts, sales engineers and managers who need shared AI vocabulary and responsible-AI concepts without building systems.
• Engineers in AWS-centric organisations where the practical AI surface is Bedrock, SageMaker and related managed services, and where the certification ladder is already understood by HR.
• Early-career people who want a broad, recognised foundational credential before specialising.
• Anyone who needs a low-stakes first exam to learn how proctored certification works.

Skip it if you already have ML background and want proof of design skill; it will feel like a vocabulary test.

Should you take both?

Often, yes, and in a sensible order:

• Building on Claude inside AWS (for example Claude models via Bedrock, or Claude Code in an AWS-heavy shop): take the CCA-F for the architecture and AWS AI Practitioner (or a higher AWS credential) for the platform. They do not overlap much, so the study does not double up.
• Not sure yet which stack you will land on: take AWS AI Practitioner first for breadth, build one small Claude project, then sit the CCA-F once you have something concrete to design against.
• Already deep in Claude, no AWS in sight: CCA-F only. Add a cloud credential later if a role demands it.

There is no conflict between them, and neither is a prerequisite for the other.

What the exams are really testing, side by side

A worked contrast. An AWS AI Practitioner-style item might ask which AWS service is appropriate for a described generative-AI use case, or which responsible-AI principle a scenario illustrates. The skill is recognition.

A CCA-F-style item describes a support agent that must never issue refunds above a threshold, offers four designs (a system-prompt rule, a PreToolUse hook that blocks the tool call, a post-hoc audit, a confidence check) and asks which is least bad given that the rule is high-consequence and easily verifiable. The skill is trade-off judgement under constraints; the answer is the programmatic hook, and the exam expects you to know why the prompt rule fails. See the study article on programmatic vs prompt enforcement.

If the second kind of question sounds like your job, the CCA-F is your exam. If the first kind sounds like a gap, start with AWS.

Next step

If you are leaning CCA-F, take the free mock exam to see whether the scenario style suits you, then start with the Domain 1 study guide. For the wider landscape of options beyond these two, read AI engineer certifications in 2026.