Sottiaux's triple: the celebration postponed (6802 likes, top of the snapshot): 'This celebration is moved to tomorrow as the button was already pressed today' — after days of teasing about finding the 20-years-unpressed reset button; plus 'Team is cooking like never before' (3372 likes) and the flight tracker 'Landing 2:30pm PST' (1877 likes) — something big seems to be landing, literally.

Nikunj's war analysis (51 likes): 'The big model labs are fighting.. this was inevitable. Watch as subsidized token prices go up in <6 months as price of compute and shortages start being "actually" felt. Say goodbye to your reset limits. Great for customer choice in the long run. And add another $1-30B in valuation to all the open harnesses in the short run 😅' Plus his anti-dunking wisdom (94 likes) and the alphabet blocks with his kid (24 likes).

Rauchg's DM philosophy (1983 likes): 'I read every DM people send me, even if I don't reply to all. I'm constantly learning about new ideas, ways to improve our product, where we're falling short, who could join our company next, where to invest… very grateful for the @x platform.' And fx 0.0.7 (238 likes): hugely improved MCP support, eval'd with top MCPs like Context7, Datadog, MongoDB, Linear, Notion, Supabase, Neon, Stagehand, Kernel; more minimalist with fewer tools and improved shell execution — 'One of the trickiest things about CLI agents is that running commands is actually not that simple. Commands might hang or run for very [long].' Zara's trio: AI product sameness (6 likes) — 'Never had so many AI products looked so similar… every time I open this app it feels like homework'; permission fatigue (0 likes) — every product asks for Gmail Calendar Notion Granola Github Slack; and the Tokyo opportunity (198 likes): 10 weeks building in Tokyo and learning the Japan AI ecosystem.

Peter Yang's trio: the weird feeling of entering 2FA into a cloud computer (42 likes); the most useful @bot — 'errands your spouse wants you to run but you delegated to a bot instead' (43 likes); and 'Cheap dad all the way' (2 likes). Nan Yu's Dune reference (13 likes): 'we're gonna need mentats aren't we.'

Matt Turck flags the RSI timeline (27 likes): 'All frontier AI labs are basically working on the same thing: recursive self improvement (RSI), meaning AI building AI. This could lead to an AI takeover as early as late 2028 or 2029, says @RyanGreenblatt.' Dan Shipper's counterpoint (17 likes): 'not an expert, but it seems like a lot of this gets solved if models don't collaborate willingly and/or are trained to report bad behavior' — plus his link of interest (45 likes): 'very interesting post and perspective.'

A re-run in today's feed: No Priors' conversation with Science founder/CEO Max Hodak (originally covered in our 2026-08-22 issue) — Prima retinal prosthesis with European marketing approval helping blind patients see again; the brain very literally is a computer (the skull IS the brain-in-a-vat); engineering beats drug discovery's card-flipping; the platonic representation hypothesis linking AI internals to neural recordings; 'I'm going to be ultimately fairly disappointed if I'm murdered by my pancreas'; the 10-bits-per-second cognitive bottleneck; and the long game of substrate independence and adapting to hard vacuum for the stars.