A slow news day in product land, but the feed is full of people thinking out loud about where AI is heading and how to act inside that uncertainty. The contrast between a VC flagging potential AI-driven civilizational risk by 2028 and a product leader joking about a bot to handle his spouse's errands is, somehow, both things that happened in the same 24-hour window.
01
Every frontier AI lab is building the same thing, and it might matter sooner than you think
FlightVC's Matt Turck flagged a sobering framing: every major AI lab right now is working on recursive self-improvement, meaning AI systems that build and improve other AI systems. He pointed to Anthropic researcher Ryan Greenblatt's estimate that this dynamic could produce an AI that meaningfully outpaces human control as early as late 2028 or 2029. ---
Why it matters: 2028 is not a distant abstraction. It's two product cycles away. If you're making 5-year infrastructure bets, hiring plans, or strategic roadmaps right now, this is the timeline your risk assessment probably isn't pricing in. Whether you believe Greenblatt's estimate or not, the fact that Anthropic's own researchers are saying it out loud is worth sitting with.
A candid note from Vercel CEO Guillermo Rauch: he reads every DM sent to him, even when he can't reply to all of them. He uses the stream as a live product feedback loop, covering gaps in Vercel's product, potential hires, and investment ideas. ---
Why it matters: For anyone building something and wondering whether cold-messaging founders actually does anything, there's your answer from one of the more active ones. Whether this scales as Vercel grows is a different question.
Peter Yang's bot idea is funnier because it's actually useful
Product leader Peter Yang floated the concept of a bot that handles the errands your spouse asks you to do, noting that the name "errands your spouse wants you to run but you delegated to a bot instead" is simply too long. It's a joke, but only partially. ---
Why it matters: The most interesting AI agent category right now might not be enterprise workflow automation. It's the small, unglamorous personal delegation tasks that people actually forget to do. Someone is going to build this seriously, and it'll be enormous.
Quick hit: Nikunj Kothari says stop dunking, start building
Nikunj Kothari posted a pointed note about the dunking reflex in tech: criticizing competitors or hyped companies feels good and gets likes, but it's ultimately a productivity trap. His advice is blunt. If you think you can do it better, go do that. --- *Note: Dan Shipper shared a link described as "very interesting post and perspective" with no additional context. There's nothing here to summarize, so it's not here.*