Carmelo Iaria
 Niharika Srivastav
Sanjay Saxena
Free Lesson

Claude Certified Architect: Safety as Architecture

Part of Claude Certified Architect: The Last Mile

30 min
Sep 30, 2026 12:00 PM

By continuing, you agree to Maven's Terms and Privacy Policy.

What you'll learn

Fail-close tools by blast radius

Match HITL and permissions to stakes. Prompt injection, write access, and compliance are design inputs.

Design the path, not a pile of agents

Pick a pattern — workflow, agentic, or augmented LLM — and decompose it so a teammate could implement it.

Put risk in the architecture, not a slide

Hazard, control, owner. GDPR, HIPAA, FedRAMP constrain the path — they are not a values appendix.

Why this topic matters

Forrester: the enterprise architect is now the human in the loop responsible for what agentic AI delivers. Labs cannot sell that. Models cannot own it. Capital is chasing people who put guardrails, HITL, and compliance on the path — not on a values slide after the demo. In 30 minutes you will fail-close the system so the organization can actually trust it. Lesson 3 of 5.

You'll learn from

Carmelo Iaria

Carmelo Iaria

Agentic Architect, Pearson SME, Startup Founder, 30 years of global experience

Cisco
Claro Brasil
Pearson
O'Reilly
 Niharika Srivastav

Niharika Srivastav

AIGP, Certified AI Governance Practitioner

Sanjay Saxena

Sanjay Saxena

Chief AI Officer, Ex-Harvard, Deloitte, KPMG

See all products from Carmelo
Get free access