Box CEO Aaron Levie is still the only person talking sense about AI agents. While everyone else argues about which model is best, he's asking who's actually going to make this stuff work in the real world.
01
Box CEO Aaron Levie: Enterprise AI agents need way more help than anyone admits
Box CEO Aaron Levie pushed back on the assumption that AI agents will seamlessly slot into existing companies. He argues that enterprises face massive infrastructure challenges — legacy tech stacks that need modernizing, fragmented data across multiple tools, undigitized knowledge, and complex change management — all while keeping their current operations running. The implication: implementing AI agents isn't a software problem, it's an organizational transformation problem.
Why it matters: If you're betting that agents will just work out of the box at your company, you're in for expensive surprises. The real money will be made by whoever solves the boring infrastructure problems, not whoever builds the smartest agent.
Microsoft Labs expanded its Pomelli experiment to European markets. Josh Woodward noted that small and medium businesses are "running with it," though he didn't specify adoption numbers or what exactly Pomelli does for these companies.
Why it matters: Microsoft is testing AI tools on European SMBs before rolling them out globally. If Pomelli works there, expect it to hit US markets with regulatory compliance already built in.
Moonshot's Kimi K2.6 aims to match Claude Opus performance
China's Moonshot AI released Kimi K2.6, positioning it as the world's leading open model that can compete with Anthropic's Claude Opus. The Latent Space newsletter flagged this as a major update that could be setting the stage for DeepSeek's upcoming v4 release.
Why it matters: Chinese AI labs are now releasing models that directly target OpenAI and Anthropic's premium tiers. Your AI budget just got more competitive, especially if you're willing to work with non-US providers.