Future cohort · Join the waitlist

Prepare for the Claude Certified Architect – Foundations (CCAR-F) exam

Work through the architecture decisions the exam tests, from agent orchestration and MCP tool design to Claude Code team setup, structured output, and reliability. The course is built for developers and architects who already build with the Claude API, the Agent SDK, Claude Code, or MCP.

Explore the three-week curriculum Join the waitlist

Live training with Mike Wheeler
AI and Salesforce trainer, O’Reilly author

The official Anthropic certification program describes the exam and its eligibility requirements. This interest list is for Velza training.

Join the waitlist

Get an email when a future cohort is scheduled. You can also tell Mike what you would like to learn to help shape the training.

Joining the interest list does not enroll you in a course. Privacy policy

Who this course is for

This cohort is for solution architects and technical practitioners with hands-on Claude development experience who want to reason through architecture and implementation tradeoffs.

The official guide describes an intended candidate who typically has six months or more of practical work across the Claude API, Agent SDK, Claude Code, and MCP. This course organizes and exercises that experience; it is not an introduction to AI chat.

Your three-week curriculum

Three weeks of live learning, with hands-on work between sessions. Join the waitlist for future cohort details.

Week 1

Design Agent Architecture and Orchestration

Turn requirements into an agent architecture that makes the agent loop, subagent roles, enforced steps, and handoffs explicit.

Map a sample support workflow into agent responsibilities
Compare coordinator and subagent designs and the context each subagent receives
Trace the agent loop and where hooks enforce a business rule
Plan enforcement points, handoffs, and when to resume or fork a session
Create an architecture diagram with responsibilities and control points

Week 2

Design Tools and Configure Claude Code

Design the tools and MCP servers an agent relies on, then decide how a team guides Claude Code, from shared instructions to the CI pipeline.

Define MCP tool contracts, structured errors, and scoped tool access
Connect MCP servers and choose the right built-in tools
Organize CLAUDE.md instructions, skills, and path-specific rules
Compare planning and execution workflows and run Claude Code in continuous integration
Create a tool-interface design and a team configuration plan

Week 3

Structure the Output and Make the Design Reliable

Specify and validate structured output, then reason about context, escalation, errors, and human review, and defend your decisions in realistic scenarios.

Use explicit criteria, examples, and schemas for document extraction
Plan validation, retries, batch processing, and multi-pass review
Manage long contexts, escalation triggers, and errors passed between agents
Plan human review, confidence calibration, and source provenance
Work original practice cases, then write a reliability checklist and decision record

What you’ll practice and keep

Architecture diagram

An architecture diagram with agent responsibilities, enforcement points, handoffs, and a session strategy.

Tool contracts and team configuration

Tool contracts for the sample agent and a written plan for shared instructions, path rules, skills, and CI review.

Output specification, reliability, and decision record

An output specification with a schema and validation, a checklist for failure and recovery paths, and a reasoned explanation of your architecture choices.

How the live course works

Three weekly live teaching sessions with worked technical examples, guided practice, discussion, and original exam-style scenarios. Bring practical development experience and use the exercises to identify gaps in your preparation. This course supplements hands-on experience; completing the course does not guarantee certification. What you need: a text editor and access to Claude. Claude Code demos are shown as recorded videos, and no API account or API spend is required.

By the end, you’ll be able to

Design agent responsibilities and orchestration paths
Define tool contracts, permissions, and escalation boundaries
Plan Claude Code team configuration and structured-output workflows
Evaluate context and reliability tradeoffs
Explain architecture decisions and identify preparation gaps

Learn through guided practice and discussion.

Exam coverage

Knowledge area	Weight
Agentic Architecture & Orchestration	27%
Tool Design & MCP Integration	18%
Claude Code Configuration & Workflows	20%
Prompt Engineering & Structured Output	20%
Context Management & Reliability	15%

Everything that’s included

Live teaching and optional office hours

Three teaching sessions and optional office hours across the three-week course. Bring questions and get help with the hands-on work.

Replay access for a full year

Revisit the recorded sessions on your own schedule as you practice and review.

Weekly hands-on projects

In week one you build an architecture diagram that shows agent responsibilities, enforcement points, handoffs, and a session strategy. In week two you write tool contracts for the sample agent and a team plan for shared instructions, path rules, skills, and CI review. In week three you write an output specification with a schema and validation, a checklist for failure and recovery paths, and a decision record that explains your architecture choices.

Readiness scorecard

Assess your understanding by exam domain and revisit your study priorities as you learn. Get the readiness scorecard.

Certificate of completion

Complete the course and receive a personalized certificate with your name, track and a public verification link. This records course completion; professional certification requires the separate certification exam.

Frequently asked questions

Is Foundations suitable for someone completely new to Claude?

This exam targets technical architects with practical Claude application experience. If you are new to using AI at work, begin with the workplace beginner course. This preparation course assumes you can discuss implementation choices and technical workflows.

Can anyone register for the certification exam?

You can join this training regardless of employer. Anthropic currently restricts exam registration to employees of Claude Partner Network organizations using a recognized company email address. Check Anthropic’s current eligibility rules before planning to take the exam. Course participation does not grant exam access.

Where can I check the official exam information?

Anthropic’s Architect Foundations certification page provides the official program information and exam guide. The weekly sequence here is Velza’s independent teaching plan.

Is there an official practice exam?

Anthropic retired its earlier practice exam when the program moved to Pearson, and the exam guide’s sample questions show the format. In class you work original practice cases built around the exam’s published domains.

What if I can’t attend a session live?

Sessions are recorded, and you keep access to the replays for a full year. You can watch and revisit the teaching on your own schedule.

Does joining the waitlist enroll me in the course?

No. Joining the list registers your interest. We will share the course details before you decide whether to enroll.

Learn with Mike Wheeler

AI and Salesforce trainer, O’Reilly author.

Learn through live instruction, practical examples, and hands-on work that connects course concepts to your own projects.

About Mike Wheeler

Stay informed about this course

Join the waitlist to receive future cohort details.

Join the waitlist