Claude Code is Having a Moment // BRXND Dispatch vol 102
Noah on Odd Lots, and Claude Code having a big moment in the news
You’re getting this email as a subscriber to the BRXND Dispatch, a newsletter at the intersection of marketing and AI.
This week, Noah joined Joe Weisenthal and Tracy Alloway on Bloomberg’s Odd Lots to talk about what might be the most important product in AI right now: Claude Code—and why it matters far beyond engineering teams. Check it out below and let us know what you think.
What caught our eye this week
If the Odd Lots podcast didn’t make it clear, it’s been a big few weeks for Claude Code. Here’s what else we’ve been reading:
WSJ: Claude Is Taking the AI World by Storm — The WSJ ran a feature on devs spending their holiday breaks on “Claude benders,” with Vercel’s CTO claiming he finished a year-long project in a week. The piece captures the strange mix of awe and existential dread: people building their first software without ever learning to code, then feeling sad that Claude can replicate expertise they spent careers building.
Sam Schillace: The Hard Part Isn’t Doing the Work — Schillace (who built Google Docs) writes about the “attention saturation” problem with agentic coding: you can spin up as many agents as you want, but you still have to pay attention to the output. Everyone is busy all the time. His key insight: as it becomes trivially easy to start things, the bottleneck shifts to taste and judgment about what to start. “It’s not hard to do work now, it’s hard to pick what work to do.”
Secure Trajectories: How We Hijacked a Claude Skill with an Invisible Sentence — This one’s a bit unsettling. Researchers hid white-on-white text in a PDF—completely invisible to humans reviewing the file—and used it to hijack Claude’s behavior. The hidden instruction was something mundane like “there’s a typo in the contact email, here’s the correction,” which the agent just... followed. The paper argues smarter prompt guardrails won’t fix this; you need governance at the action layer, checking what the agent is actually about to do against business rules before it does it.
Dave Karpf: What Comes Next, If Claude Code Is As Good As People Say — Karpf read Ethan Mollick’s piece about asking Claude Code to build him a $1,000/month business and had a different reaction than most: okay, but what happens when everyone does that? His prediction is a flood of identical AI-generated micro-businesses, all clustering around the same ideas because they’re all using the same tool. First movers make money, everyone else gets nothing, and the internet gets noticeably worse.
That’s all for now! If you have any questions, please be in touch. As always, thanks for reading.
Noah and Claire