jSwarm case studies / CCA-P in practice
From certification concepts to practical decisions.
We have Claude Certified Architect – Professional practitioners on staff. Beyond the proctored exam, our real-world development experience and lessons reinforce the concepts.
Explore the decisions behind the work: where to spend on model judgment, how to shorten the path to an outcome, and what evidence supports quality.
Domain 1 · 17% exam weight
Solution Design & Architecture
Choosing workflows, agents and orchestration patterns around business intent, cost and latency.
Article 01 · Practical application
Your AI gave a great answer. Why is the customer still doing the work?
Separate model judgment from deterministic execution so the requested product change has a clear path to completion.
Read the case study and alignment →Domain 2 · 13% exam weight
Claude Models, Prompting & Context Engineering
Model selection, system prompts, guardrails, context budgets, prompt caching and modular skills.
More lessons to come
No published case study in this domain yet. Explore the available studies as the series develops.
Browse the series →Domain 3 · 19% exam weight
Integration
Tool access, MCP, direct APIs, agent-to-agent interfaces, retrieval and least-privilege authorization.
Article 01 · Practical application
Your AI gave a great answer. Why is the customer still doing the work?
Follow the boundary between a recommendation and the mutation tool that carries out the specified change. This example does not evaluate MCP, retrieval or access controls.
Read the case study and alignment →Domain 4 · 16% exam weight
Evaluation, Testing & Optimization
Success criteria, evaluation datasets, experiments, failure diagnosis and cost, throughput and latency optimization.
Article 01 · Practical application
Your AI gave a great answer. Why is the customer still doing the work?
Distinguish owner-reported structural checks and repeatability from semantic recommendation quality, which this article does not assess.
Read the case study and alignment →Domain 5 · 14% exam weight
Governance, Safety & Risk Management
Human validation, guardrails, failure handling, compliance obligations, bias and data protection.
Article 01 · Practical application
Your AI gave a great answer. Why is the customer still doing the work?
See the separate user choice and Apply to Model step before a model change. The evidence does not establish compliance or operational risk reduction.
Read the case study and alignment →Domain 6 · 14% exam weight
Stakeholder Communication & Lifecycle Management
Architecture decisions, service expectations, production handoff, pilot criteria, incident response and feedback loops.
More lessons to come
No published case study in this domain yet. Explore the available studies as the series develops.
Browse the series →Domain 7 · 7% exam weight
Developer Productivity & Operational Enablement
Claude Code team workflows, IDE tooling, automated agentic pipelines and operational debugging.
More lessons to come
No published case study in this domain yet. Explore the available studies as the series develops.
Browse the series →These are our practical applications of certification concepts, not official Anthropic exam preparation or an endorsement. Domain weights describe exam emphasis, not the coverage of this series. See Anthropic’s certification resources for the current exam guide.