Sam Altman's most-engaged tweet of the day (11071 likes, 294 retweets): 'our older kid put two words together for the first time and i am approximately as amazed by this cognitive feat as i am by GPT-5.6 discovering new math' — a humanizing moment from the OpenAI CEO that resonated massively.

Rauchg made a bold World Cup prediction (1545 likes, 310 replies): '🇺🇸 USA - 🇦🇷 ARG final. You heard it here first.' Amjad Masad celebrated Independence Day: 'Happy 250 🇺🇸' (791 likes). Matt Turck praised Norway: 'best jersey, best fan celebration, best striker' (70 likes) and noted this is a 'tournament of the older goalies' (64 likes).

Cat Wu shared her new favorite Claude Code workflow for sourcing candidates (555 likes, 92 replies): tell Claude Code about the role and backgrounds you want → it kicks off a dynamic workflow to find 100 candidates with LinkedIn, Twitter, blog, podcast links and a one-line pitch for each → ask CC to make an artifact and email it to you → lock your laptop and head out → review the list on the go once CC is done. End-to-end agentic recruiting.

Nan Yu (Head of Product at Linear) with two sharp takes: (1) 'Bragging about running 10 Claude code tabs is just theater' (12 likes) — pushing back on performatively complex setups. (2) 'The real-time strategy game model of managing agents is clearly a dead end. AI that's ancient by current standards already beats >99% of human players and out-micros them to an extreme degree' (31 likes) — arguing that humans micromanaging agents is a losing strategy.

Thariq's SF joke (581 likes, 43 replies): '"God gave me a sign" (remembering i live in San Francisco) "I mean, I was acausally influenced by the ASI at the end of time to maximize EV for humanity"' — a perfect parody of how rationalist SFculture reinterprets everyday experiences.

Dan Shipper's Fable parody (153 likes): 'me: change this button color. Fable: sure I just spun up a fleet of 100 agents to get that done for you' — capturing the feeling of disproportionate computational effort behind seemingly simple requests.

Amanda Askell (Anthropic philosopher) on a universal frustration (1382 likes, 131 replies): 'Extracting a probability from a doctor is one of life's unnecessary boss battles. Even if you beg them for an interval-valued subjective probability, and at that point you're basically asking for their hunch. I don't know if they get sued for giving out information or something.'

Garry Tan, from Osaka, with a philosophical pair of tweets (51 + 39 likes): 'I'm in Osaka now, and it's clear: Japan already ran the experiment on the ceiling. Thirty years of zero growth and it built the best trains, service, and craft on earth. When you can't compete on more, you compete on better.' Followed by: 'So the real constraint on human wealth was never resources. It was good ideas for how to serve one another, and the leverage to act on them. We just deleted the leverage constraint for everybody. Now it's only the ideas. Go have them, and then build them.'

Nikunj's VC meeting rant (113 likes, 16 replies): 'Still waiting on that one founder who gates a VC chat until they have played with the product and come with at least two pieces of feedback. Blows my mind that people go on Zoom and chat for 30 mins on a parroted deck... It's just more efficient for both sides to upload their personal "prompts" to Claude at this point 😂'

Zara resurfaced her code-understanding skill (125 likes): 'Resurfacing this skill I built (now that understanding your code is in vogue)' — timed to the wave of interest in code comprehension tools. Steipete recommended something with just 'Can't recommend enough' (184 likes).

Cloudflare CEO Matthew Prince delivers a masterclass on the state of the internet. Key milestone: in H1 2026, bot traffic surpassed human traffic online for the first time. He predicts in 5 years there will be 1000x more bot traffic than human traffic — agents visit 5000 sites where a human would visit 5.

The internet's business model (advertising) is dead: 'bots don't click on ads.' The next business model is undefined. Cloudflare is working on micro-payment infrastructure that would need to handle 10M transactions/second at launch (Visa does <100K/s), using protocols like x402.

On the layoff: Cloudflare cut 20%+ of its team — not because business was struggling, but because AI tools eliminated the need for many middle management roles. Span of control is moving from 6:1 (Harvard B School standard) to 12:1. 'The cruelest thing you can do is wait.' He expects most companies will do the same in 6-12 months.

On security: for the next 2 years, we'll see a Log4j-scale vulnerability every week as AI finds them all. But then software will get dramatically better. Cloudflare's internal agent — trained on 10 years of incidents — already reduced their background incident rate off a cliff.

On Cloudflare OS: an internal AI platform for non-engineering teams (legal, finance, marketing). The trick to capturing real workflows: set up a 'magic AI agent' email address (autocloudflare.com) staffed by 20 humans, gather real problems, then build skill files. IR document prep went from 2 weeks to 3 minutes.

On the future of content: Prince bought his local Park City newspaper and expects AI licensing deals to exceed ad revenue this year. He envisions a Spotify-like model where AI labs reward creators who fill 'holes in the Swiss cheese' of human knowledge — net new information, not another story about 1500 Pennsylvania Avenue.