Test your Professional readiness

Sample Claude Certified Architect Professional practice questions

6 free sample questions from the practice bank, with the correct answer and the reasoning.

1. Question 1 · Domain 1 — Solution Design & Architecture

   Logistics Quote System

   A logistics company wants Claude to process inbound freight quotes. Every request follows the same steps: extract shipment details from an email, validate them against a rate card, and generate a quote document. Requirements are stable and the steps never vary. Which architectural pattern is most appropriate?

   • A.A multi-agent system with a supervisor agent delegating to specialist agents
   • B.A fixed workflow where each step is a discrete, sequenced LLM call Correct answer
   • C.An autonomous agent with tool access that plans its own approach per request
   • D.A single monolithic prompt containing all instructions and the full rate card

   Why: The task is stable, repeatable, and fully known in advance — the textbook case for a fixed workflow, which gives predictability, per-step auditability, and easy debugging. A adds multi-agent coordination overhead the problem doesn't need. C adds autonomous planning overhead the problem doesn't need. D sacrifices step-level control and validation.

2. Question 2 · Domain 2 — Claude Models, Prompting & Context Engineering

   High-Volume Classifier

   A ticket-routing system classifies 400,000 short messages per day into eight categories. Accuracy on a labelled test set is comparable across the model family, and the business is sensitive to both latency and cost. Which model strategy is most defensible?

   • A.Use the most capable model, since classification errors are always more expensive than compute
   • B.Use the mid-tier model as a compromise, avoiding the need for evaluation
   • C.Use the smallest model that meets the accuracy target, and validate the choice with ongoing evaluation Correct answer
   • D.Alternate randomly between models to average out their weaknesses

   Why: When accuracy is comparable across models, the disciplined choice at 400k requests/day is the smallest model that meets the target — provided ongoing evaluation confirms it keeps meeting it. A substitutes assumption for measurement. B avoids the evaluation that should drive the choice. D makes behaviour unpredictable and undebuggable.

3. Question 3 · Domain 3 — Integration

   Enterprise System Connectivity

   An enterprise is connecting Claude to twelve internal systems (CRM, ticketing, HR, inventory, and others). Different teams own each system, tools will be added and retired frequently, and multiple AI applications across the company will need to reuse the same connections. Which integration approach best fits these requirements?

   • A.Hard-code each system's REST API directly into each application's tool definitions
   • B.Build one custom middleware service that wraps all twelve systems behind a single endpoint
   • C.Expose each system via MCP servers that any application can connect to Correct answer
   • D.Grant Claude direct database access to each system to avoid API maintenance

   Why: MCP is designed for exactly this shape: many systems, many consuming applications, decentralised ownership, and frequent tool churn — each team maintains its server once and every application reuses it. A multiplies maintenance by application count. B creates a bottleneck team and single point of failure. D bypasses business logic and access control.

4. Question 4 · Domain 4 — Evaluation, Testing & Optimization

   Contract Review Launch Gate

   A steering committee asks whether a new contract-review assistant is 'good enough to launch.' The team has anecdotes from pilot users but no formal measures. What should the architect establish first?

   • A.A general benchmark score for the underlying model, published by a third party
   • B.A count of pilot users who described the assistant as helpful
   • C.A comparison of the assistant's token cost against the incumbent manual process
   • D.Task-specific evaluation metrics tied to business outcomes (e.g. clause-extraction accuracy, review-time reduction, escalation rate) with agreed launch thresholds Correct answer

   Why: 'Good enough' only means something against defined, task-specific metrics with agreed thresholds tied to the business outcome. That definition must precede the launch decision. A measures the model, not the solution. B is anecdote. C measures cost, not quality.

5. Question 5 · Domain 5 — Governance, Safety & Risk Management

   Procurement Agent

   An agent can draft supplier emails, update internal records, and issue purchase orders up to $50,000. The business wants to add human oversight without slowing every interaction. Where should the human-in-the-loop approval gate be placed?

   • A.Before every action the agent takes, so nothing occurs without approval
   • B.After the purchase order is issued, so a human can review it retrospectively
   • C.Before irreversible or high-impact external actions (issuing purchase orders), while allowing low-risk drafting and internal updates to proceed automatically Correct answer
   • D.Only on the first ten requests each day, as a statistical sample

   Why: Human-in-the-loop is a targeted control: place the gate where actions are irreversible or high-impact (financial commitments leaving the building), and let low-risk steps flow. A destroys the efficiency case. B reviews after the money is spent. D leaves most high-impact actions ungated.

6. Question 6 · Domain 6 — Stakeholder Communication & Lifecycle Management

   '100% Accuracy' Demand

   During discovery, a client's executive sponsor insists the AI solution must be '100% accurate' before launch. As the architect, what is the most effective response?

   • A.Agree to the requirement and add extended testing phases until accuracy reaches 100%
   • B.Explain that LLM systems are probabilistic, then work with the sponsor to define measurable acceptance criteria and an error-handling strategy aligned to business risk Correct answer
   • C.Decline the engagement, since the requirement cannot be met
   • D.Propose a smaller model to reduce the risk of errors

   Why: The architect's job is to convert an unachievable absolute into an achievable agreement: probabilistic systems, measurable acceptance criteria, and error handling proportionate to business risk. A commits to the impossible. C walks away from a solvable conversation. D changes the error rate, not the expectation.

What this practice exam covers

554 questions · 7 exam domains · 2 mock exams

• Domain 1 — Solution Design & Architecture · weight 17%
• Domain 2 — Claude Models, Prompting & Context Engineering · weight 13%
• Domain 3 — Integration · weight 19%
• Domain 4 — Evaluation, Testing & Optimization · weight 16%
• Domain 5 — Governance, Safety & Risk Management · weight 14%
• Domain 6 — Stakeholder Communication & Lifecycle Management · weight 14%
• Domain 7 — Developer Productivity & Operational Enablement · weight 7%

Questions about this practice exam

Are these the official Claude Certified Architect Professional exam questions?

No. They are practice questions written against the published exam domains. This site is not affiliated with or endorsed by Anthropic.

How many practice questions are there?

554 questions across 7 exam domains including 2 full mock exams.

Do I need an account to practise?

No. You can take the practice questions without an account and your scores save in your browser. A free account adds the full answer review, gap analysis and a completion certificate.

Do the questions explain the answers?

Yes. Every question comes with an explanation of the correct answer.