Anthropic Engineers Are Using A New Claude Code Prompting Trick...

The head of product at Anthropic just shared a simple trick that makes Claude compete against itself until it hands you the best result possible. ⚡ Join the Dream Labs Community: https://app.joindreamlabs.com/join?v=... In this video we put Angela Jiang's tournament strategy to the test live, spinning up three independent Claude sub-agents to build the same landing page in isolation, then having a blind judge pick the winner. We compare it side by side against a standard one-shot prompt so you can see exactly what the difference looks like and whether it's actually worth it. In this video we cover: 🏆 What compete mode is and how Angela Jiang described it 🧪 The live test: control vs tournament competition 🎯 How the blind judging works and what scores each contestant got 🌐 The winning page compared to the control side by side 📊 Where this strategy ranks against loops, context and omnipresent AI ⏱️ Timestamps: 0:00 Intro 0:40 Angela's Strategy 1:46 Setting Up the Control 2:23 Setting Up the Tournament Competition 4:07 Control Results 5:25 Tournament Results and Blind Judge Scores 6:02 The Winning Page Compared 7:23 Our Verdict and Where This Ranks #ai #claudecode #anthropic #aitools #aiautomation #artificialintelligence #productivity #nocode #aiforbeginners #business