The agent hiring paradox is here: companies are adopting AI agents and hiring more humans at the same time. The reason why tells you everything about where this technology actually stands.
01
Box CEO Aaron Levie: AI agents are creating jobs, not eliminating them
Aaron Levie shared insights from his enterprise conversations outside Silicon Valley: most companies are hiring while adopting AI agents. There's huge demand for technical talent to build software and act as "field development engineers" for agents. As AI handles routine tasks in customer workflows, companies are doubling down on client-facing roles that need human judgment.
Why it matters: The "AI will replace everyone" narrative is missing what's happening in real companies. Someone has to build, monitor, and fix the agents when they break. If you're in engineering or customer success, this wave is creating opportunities, not eliminating them.
FirstMark's Matt Turck offered a contrarian take: what if AI just makes us more productive without changing everything? He called it "the biggest mindf*ck scenario" where both doomers and accelerationists are wrong. We get enterprise automation, some scientific breakthroughs, better productivity. That's it.
Why it matters: Most AI investment is betting on revolutionary change. If Turck is right, a lot of $100 billion valuations are based on science fiction, not business fundamentals.
Zara Zhang detailed her shift in coding workflow over the past month: she moved from terminal to desktop apps (especially Codex's Mac app), split her usage 50-50 between Codex and Claude Code, and stopped opening the terminal entirely. She uses Codex for defined tasks that just need to work, Claude Code for unclear requirements that need planning.
Why it matters: This is what the coding agent market actually looks like in practice. Developers aren't picking one tool and sticking with it. They're building workflows around different AI personalities for different types of problems.
OpenAI featured Warp's approach to coordinating coding agents across local, cloud, and open-source development using GPT-5.5. The terminal company is using OpenAI's models to manage complex development workflows that span multiple environments.
Why it matters: This is OpenAI's answer to Google's coding agent demos. Instead of showing off lab experiments, they're highlighting real products that developers actually use in production.
AI infrastructure investor Swyx posted a simple observation: "ai infra is going VERTICAL." Short on details but the timing suggests he's seeing specialized AI infrastructure companies focusing on specific industries rather than building general-purpose tools.