Skip to content
Anthony Wang
Illustrated avatar of Anthony WangPrincipal Architect · Google Cloud APAC

Enterprise AI succeeds or fails in the gap between demo and production.

> 

Anthony Wang is Principal Architect at Google Cloud APAC — fifteen years across engineering, presales, and enterprise architecture, advising boards and C-suites of the region's largest conglomerates, manufacturers, and automotive groups on agentic AI and digital transformation, alongside global system integrators.

Today's AI briefingRead the writing
Career
Google CloudSiemensBoschSAP
Credentials
TOGAFGoogle Cloud PCANUS Computer Engineering

01 // Daily briefing

The daily AI briefing

All briefings
Daily Briefing

AI Briefing — 27 July 2026

Nvidia and SK Group sign a $500B+ Korea infrastructure pact, Chinese open-weight models now handle 58% of US firms' routed tokens as Beijing weighs new export curbs, and a Japanese automaker commits to mass-producing humanoids.

  1. 01Top stories
  2. 02My take
Read the briefing

02 // Latest writing

Latest writing

View all
Deep Dive

ADK 2.0's Workflow Runtime: deterministic graphs for the steps agents shouldn't improvise

ADK 2.0 swaps the hierarchical agent executor for a graph-based Workflow Runtime: deterministic nodes, first-class human-in-the-loop, and a benchmarked case for using them.

  • adk
  • agents
  • workflow-runtime
  • google-cloud
Perspective

Your agent tool layer just became infrastructure

MCP's July 28 revision deletes protocol sessions, hardens auth around the enterprise identity stack, and designs explicitly for gateways. Connector-building is ending as differentiating work — the hard question is now what your agents may reach, and who decides.

  • mcp
  • agents
  • governance
  • platform
Deep Dive

Evaluating multi-agent systems: a practical loop with Google's ADK

A working evaluation loop for multi-agent systems built on Google's Agent Development Kit — datasets, deterministic checks, LLM judges, and the failure modes that actually show up in production.

  • adk
  • agents
  • evaluation

03 // Technical assets

Reference architectures & code

View all
Code

ADK eval harness starter

A minimal, extensible evaluation harness for agents built on Google's Agent Development Kit — run a dataset against an agent, capture traces, and score outcomes.

  • adk
  • agents
  • evaluation
  • python
Architecture

Reference architecture: enterprise agentic AI platform

A layered reference architecture for running agentic AI in a large enterprise — channels, agent orchestration, grounding, and the governance plane that makes it production-viable.

  • agents
  • architecture
  • google-cloud