Claude's launch (4249 likes, top of the day): 'You can now discover tools, agents and expert partners on Claude Marketplace. Use it to: Add connectors and plugins like Slack and Notion; Buy agents and products from companies like Cursor and CrowdStrike; Scale with service partners like Accenture and Deloitte' — plus the marketplace browse link (179). Rauchg's double: the anatomy lesson (1453): 'Muse, Instinct, OpenClaw, Claude Code… All successful agents have 3 key components: 🧠 Brain → model, harness (logic); 👐 Hands → tools, computer, browser; 🗃️ Files → memories, skills, repos. The "easy" way is to throw all these in 1 stateful computer (a Mac Mini)… But if you want to cost-efficiently run agents in the cloud…'; and the ShellPerfBench tip (112): 'I'm really neurotic about the startup time of a new shell session. Opus 5.5 found a lot of really great optimizations other models missed. Anon, anytime you open a new terminal, you might be silently suffering. Tell your agent to optimize .zshrc and friends.' Garry Tan's trio: 'Two important simultaneous trends in how startups get customers: Make software agents want. Use agents to make people want software' (434); 'This is not priced in yet' (428); and 'Muse is very impressive' (292).

Sottiaux's double: Voice! (4009 likes): 'I've gotten so used to just call ChatGPT and talk about all the work, check on emails, do some coding, manage my calendar and it works across the full ecosystem of plugins, including 3P developed plugins'; and the DevDay tease (3159): 'Can't wait for DevDay next Tuesday. Some really fun stuff, but also many many things that should change the way you work. It's been our most ambitious sprint and Astra has really made new things possible in such short amounts of time.' Cherny's trio: the speed post (3292): 'If you've noticed how fast claude.ai and the Desktop app have become in the last few weeks, here's how we did it. Lots of juicy learnings & techniques in the blog post'; meeting Knuth (1417): 'It was an honor to meet Donald Knuth at the Computer History Museum today'; and the formal-methods footnote (68): 'Claude is doing something like: 1. Building a model of the program… 2. Finding counter-examples… 3. Reproducing the bugs 4. Fixing the bugs. It's not that the whole codebase is formally verified (yet!)'. Swyx's AINews verdict (205): 'ran @latentspacepod AINews side by side with 6 Sol and the difference was night and day. 5.5 Opus is the new default model for AINews going forward. So much more concise and tasteful reporting, with much less slopese than even 5 Opus.' Peter Yang's trio: the Opus 5.5 list (309): 'Smart / Fast / Fun to talk to / Good at writing / Friendly on rate limits / Constantly surprising'; the slides prompt (206); and the harness gap (25): 'With a model this good the Claude Code harness really needs great live voice and browser + computer use. Right now one lab has the best model and another has the best harness.'

Thariq's trio: the new format (2289 likes): 'this is a new type of post we're trying where we share in-depth about how we do work using specific prompts and techniques you should be able to replicate. let us know if it's useful!'; the reality check (1265): 'the post: "Claude one-shot this". the prompt: 10k characters with good takes plus skills, examples and API keys'; and 'also it does look incredible' (101). Ryo Lu's manifesto (1087): 'efficiency, productivity, speed — we keep worshipping these words like they are what we live for… but what is the point of endless production if there is no time left to think deeply? not react, not optimize, not summarize, not clear the queue — think. to sit with a question long enough that your real intention starts to appear.' Madhu's counter (146): 'I respectfully disagree with @benthompson on this. Consumers don't have one relationship with "doing things". Browsing clothes can be entertainment (for some). But hiring a roofer is miserable (for most). Even with shopping, sometimes you want to browse for an hour. Sometimes you want the right thing just delivered.' Aditya's double: 'You think Zuck is good when he is down and in the trenches and in war mode? Just see how he dominates when he has the big stack. Proud Facebook alum. LFG.' (409); and the team metric (136): 'A team that has been working together for 3+ years is significantly higher throughput and more resilient… A key mistake… is to simply ask about team size as opposed to team longevity. AUC for teams is the right metric.'

Levie's double: 'Muse and Box, together at last' (186); and the creative-industry vision (81): 'As the barriers and the costs come down, more films will get made, not fewer. Studios will get to take more risks… In the 1980s, animation was dismissed as a niche corner of the business. Today it is one of the most beloved and profitable.' Shipper's experiment (31): 'tomorrow at 6pm we find out if an ai agent can throw a good party. the @every agent planned every's september meetup from the food and drinks menu to the guest list. come grade it!' — plus 'sick' (264). Woodward's Gemini apps rollout (366): 'More of your favorite apps are coming to Gemini! Which ones do you want next?' Steipete's trio (96/89/14): 'Astra is crushing it'; 'Who needs VNC when there's browsers'; and 'crushed it'. Amjad's double: 'You need young people to break the mold, question deeply held beliefs, and attempt the impossible. The ideal college would cultivate exactly that' (284); and the eyes emoji (88). Nikunj's pair (50/15), plus Swyx's stragglers (12/4).