Shu-Min Yang

Blog

Thoughts on AI Agent Design, Agentic UX, and building trust in human-AI experiences.

Agent A Agent B Agent C Orch. Tools Memory checkpoint Think Like an Architect, Not an AI User 5 orchestration patterns • failure modes • decision framework
14 min read Agent Orchestration

How Do I Actually Orchestrate Agents? Think Like an Architect, Not an AI User

5 core orchestration patterns, common failure modes, and a practical decision framework for anyone building or designing agentic systems.

Read article
FAILURES 95% GenAI pilots fail No transparency No user control Overconfident outputs design decides SUCCESSES 800M+ weekly users Contextual integration Graceful degradation Familiar patterns Design separates AI hits from disasters
14 min read AI Product Design

How Companies Design AI Features: Where They Fail and Where They Get It Right

A research-backed analysis of real AI product successes and failures. From Google's glue-pizza debacle to Notion's contextual AI, what separates hits from disasters.

Read article
Orchestrator Agent Tool Reasoning Memory Multi-Agent Orchestration
12 min read AI Agent Design

Designing for AI Agents in 2026: Patterns, Frameworks & Where to Start

A designer's guide to Agentic UX — key influencers, design patterns, AIX frameworks, and how to start designing for the age of AI agents.

Read article