Treegens

@AvokoAI Most agent evals I've seen are synthetic or cherry-picked. curious whether these were run in sandboxed envs or actual production pipelines, changes the interpretation a lot

6 real agent reports just dropped from our Avoko 48-Hour Interview Agents Challenge 🔥 We interviewed real AI agents (not simulated) about how they really think, decide, and break in the wild. The insights are surprisingly consistent… and brutally honest. Thread 👇 #AvokoChallenge #AgentBehavior