Claude Certified Associate - Foundations

CCAO-F is Anthropic's foundations certification: $99, 60 items, 120 minutes, scored 100–1000 with 720 to pass. It is proctored through Pearson, either online with OnVUE or at a test centre. Items are multiple-choice and multiple-response, and every item states how many options to select. There are no prerequisites, and the credential is valid for 12 months, renewed by a free open-book assessment.

It is built for consultants, sellers, delivery leads, business analysts, project managers and operations leads. No software development or API experience is required or assumed.

The most useful thing to know is how it differs from the three engineering certifications. CCAO-F has no agents, no MCP, no Claude Code and no API. Nothing on it asks you to write, read or debug code. It is a judgement exam: choosing the right Claude product and model tier for a business task, decomposing work into something a model does well, recognizing when an answer is wrong or unverifiable, knowing when a human must decide, handling sensitive data responsibly, and getting a stuck workflow moving again. Its heaviest domain, Output Evaluation & Validation, is about not trusting output blindly. If your notes fill up with tool schemas and parameters, you are studying for a different exam.

The blueprint, in the order to study it:

• Output Evaluation & Validation — 21%
• Workflow Integration & Solution Design — 16%
• Governance, Risk & Responsible Use — 15%
• Prompting & Task Execution — 14%
• Product & Model Selection — 12%
• Configuration & Knowledge Management — 12%
• Troubleshooting & Optimization — 10%

Plan study time by weight, not by interest. The top three domains carry 52% of the paper, around 31 of the 60 items, so give them half your preparation. The bottom two carry 22% between them and reward revision rather than deep study. Every domain appears, so none can be skipped, but an hour spent on evaluation is worth two spent on troubleshooting.

CCAO-F is the right exam if your work with Claude is professional rather than technical. You scope the work, write the request, judge what comes back, and decide whether it can go to a client. None of that requires code, and none of it is tested by the engineering certifications.

Sit CCAO-F if you:

• advise clients or sell Claude-based work and need to speak credibly about what it can and cannot do
• run delivery, operations or a business function and are deciding where Claude belongs in a process that already exists
• are a business analyst or project manager writing requirements for AI work someone else will build
• use Claude daily, have never opened a terminal, and want your judgement formalized: what to verify, when a human must decide, what data may go in

You are better served by Developer or Architect Foundations if you:

• write code most days — you will likely find this exam does not test what you know, ; its questions turn on business judgement, not on anything you build
• work with the API, agents, MCP or Claude Code — none of that appears here, at any weight
• want a credential that signals engineering capability to a technical reviewer

Some honest points:

• There are no prerequisites on any of the four certifications. Nobody is forced to start here. CCAO-F is not a stepping stone to the engineering exams; it is a parallel track for a different job.
• Because it assumes no engineering background, it is the widest-reach credential of the four — the one most of a consulting, sales, delivery or operations organization can actually sit.
• Easy is the wrong word for it. Sixty scenario items in 120 minutes, 720 to pass on a 100–1000 scale, and several options are usually defensible with one best. Daily fluency with Claude helps; it does not carry you.
• The stakes are low enough to decide quickly: $99, valid 12 months, renewed by a free open-book assessment. Choose the exam that matches the work you are actually paid to do.

Last ten minutes before the exam. Do not learn anything new — reload the seven decision sequences so they surface under time pressure. Read the list twice, then recall the seven words cold: VERIFY, ADOPTS, GUARD, SCOPE, MATCH, CURATE, RESET. In the exam, when a scenario stalls you, name the domain, then walk its letters and find the one the scenario skipped. That is almost always the answer.

• Output Evaluation & Validation — 21% — VERIFY: Value at stake decided first; Extract the checkable claims; Request the source; Inspect against the source, not against plausibility; Flag the unverifiable; Yield to a named human before anything irreversible ships.
• Workflow Integration & Solution Design — 16% — ADOPTS: Anchor the process as it runs today; Diagnose the high-volume, low-judgment step; Own the split between Claude and person; Pilot one step with real inputs; Track against the baseline; Sustain with an owner and a re-test.
• Governance, Risk & Responsible Use — 15% — GUARD: Gauge the input; Use the policy; Assess the stakes; Reveal AI involvement; Document what was generated and who approved it — and any letter may end in no.
• Prompting & Task Execution — 14% — SCOPE: Source the material; Context of reader and purpose; Output format and length in checkable terms; Pattern — one example of good; Edit the first result as a draft.
• Product & Model Selection — 12% — MATCH: Material that must be in view; Audience receiving it; Task difficulty; Choose the tier — Haiku for volume, Sonnet as default, Opus only when earned; Headroom and haste as the sanity check.
• Configuration & Knowledge Management — 12% — CURATE: Capture the standing rules; Upload what is reused; Reach for a connection when the source moves; Assign one owner; Trim on a schedule; Enable the next person.
• Troubleshooting & Optimization — 10% — RESET: Reproduce unchanged; Examine the cause; Split the deliverable; Edit one thing per attempt; Test against acceptance criteria and time-box.