Leadership AI Model Routing: What Leaders Need to Decide Before Every Task Hits the Frontier Model
Cisco routed 90,000 employees to 800+ subagents, sending most requests to cheap models. Here is the leadership decision that made it work.
Latest
View all articles →Shell C3 AI Reliability: How Agentic AI Moves From Predicting Equipment Failures to Diagnosing Their Cause
Shell expanded its eight-year C3 AI reliability program in 2026 to diagnose root causes and recommend fixes, not just flag failures.
HealthcareKaiser Permanente Ambient AI Scribes: How 24,000 Physicians Cut Documentation Time by 15,700 Hours
Kaiser Permanente deployed Abridge ambient AI scribes to 24,000+ physicians across 40 hospitals, the largest healthcare gen AI rollout, with independently published outcomes.
LeadershipBuild vs. Buy AI Models: What Leaders Need to Decide Before Funding a Proprietary LLM
Thomson Reuters spent $40 million building its own legal AI model, then kept using Claude for CoCounsel anyway, a build-vs-buy framework leaders can copy.
RetailHome Depot AI Phone Agents: How Gemini Enterprise Cuts Call Resolution to Four Times Faster
Home Depot piloted Google Cloud AI voice agents in 50 stores, cutting phone support to four times faster while keeping a human option live.
Financial ServicesAmerican Express Fraud Detection: How AI Screens Billions of Transactions in Two Milliseconds
American Express built an NVIDIA-powered deep learning system that screens billions of card transactions for fraud in roughly two milliseconds apiece.