Shafkat Sakeeb

We're shipping a new feature in Claude Cowork as a research preview that I'm excited about: Dispatch! One persistent conversation with Claude that runs on your computer. Message it from your phone. Come back to finished work. To try it out, download Claude Desktop, then pair your phone.

@addyosmani A 5-min setup and usage with open claw and Claude code/codex x.com/misbahsy/statu…

@ProofofMaro So romantic

Topology is my love language

Cardiologist wins 3rd place at Anthropic's hackathon. Out of 13,000 applications. Built in 7 days by Michał Nedoszytko MD. Coded day and night - in the hospital, in the cloud, while flying from Brussels to San Francisco. A few years ago, it would have been impossible for a doctor to build this alone in just a couple of days. AI changed that. The project is called postvisit.ai. It is an AI agentic care platform for patients. Including reverse AI scribe it is a companion that guides the patient from the moment they leave the doctor's office. Powered by the massive context window of Opus 4.6, it allows patients to explore their full medical history, connected devices, Evidence Based resources and external data sources — all in one place. Today, the barrier to entry has vanished; even a practicing physician can build an application from scratch.

I was very, very young when I learned about this patent. Haven’t watched television since.

@internetvin Living in Montreal, what’s up?

Please reply to this if you live in Toronto or any city that surrounds it and you’re interested in clawdbot, moltbot, openclaw or whatever you want to call it and you want to explore different use cases, workflows and potentials Thank you for everything.

@hugomercierooo This is great! Would love to give it a try for our company!

𝗧𝘄𝗶𝗻 𝗶𝘀 𝗻𝗼𝘄 𝗮𝘃𝗮𝗶𝗹𝗮𝗯𝗹𝗲 𝘁𝗼 𝗲𝘃𝗲𝗿𝘆𝗼𝗻𝗲. twin.so We can’t wait to see what you’ll build. 𝗥𝗧 𝗮𝗻𝗱 𝗰𝗼𝗺𝗺𝗲𝗻𝘁 “𝗧𝘄𝗶𝗻” 𝘁𝗼 𝗴𝗲𝘁 𝗳𝗿𝗲𝗲 𝗰𝗿𝗲𝗱𝗶𝘁𝘀 𝗶𝗻 𝘆𝗼𝘂𝗿 𝗗𝗠𝘀 👇 twitter.com/hugomercierooo…

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗧𝘄𝗶𝗻 — 𝘁𝗵𝗲 𝗔𝗜 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗯𝘂𝗶𝗹𝗱𝗲𝗿. No setup. Secure. Infinitely scalable. We just raised a $𝟭𝟬𝗠 𝘀𝗲𝗲𝗱. After a beta with 𝟭𝟬𝟬,𝟬𝟬𝟬+ 𝗮𝗴𝗲𝗻𝘁𝘀 𝗱𝗲𝗽𝗹𝗼𝘆𝗲𝗱, we’re now opening to everyone. RT and comment “Twin” — first agents on us. 👇

@rachelclif Exactly why the West failed and birth rates are in decline

the markers of a great relationship: 1. the sex is great 2. the companionship is great if both these things are true, you’re onto a winner

@adamdotdev Being fast and snappy kills being smart any day

It’s wild to me how many people don’t realize how much smarter 5.2 is than opus, it’s not even close. My guess is this is entirely based on the Claude max plan token subsidy

What a brilliantly written and honest article

The coupling is coming. Too little → you're outcompeted by those who coupled more Too much → you lose capacities you can't recover There's no opting out. Only navigating consciously. Cowork is the beachhead. Claude OS is the destination. The question: how deliberately will you get there?

All of this requires deep access to your digital life. The trust stack that makes it safe: → Granular permissions (read/write/execute, per-folder, per-agent) → Audit logs and rollback for everything → Approval tiers from auto-execute to require-approval → Kill switches at every level Delegation without abdication.

Claude Cowork is just the beginning. The trajectory is clear: Cowork → Claude OS A persistent AI layer beneath your entire digital life. Not a tool you invoke. A cognitive symbiont that's always running. Here's what that actually means:

@Motion_Viz 2026

For the people who missed the workflow: Still the best workflow. here's my gift: the entire AI workflow I used to ship 5+ projects last month. most people use AI tools separately. I chained them into one pipeline. the stack: → NotebookLM — idea mapping, research synthesis → Gemini Gems — requirements extraction, analysis → Claude Code (Opus 4.5) — PRD, documentation, system prompts → Antigravity — free website builds (buggy but ships fast) the flow: raw idea ↓ NotebookLM (connects dots you'd miss) ↓ Gemini Gems (research → actionable requirements) ↓ Claude Code (requirements → production-ready PRD) ↓ Antigravity (PRD → working prototype) ↓ live product from foggy idea to shipped product in hours. this isn't using AI. this is orchestrating AI. includes: – full 12-page workflow guide – frontend system prompt template – backend system prompt template – stage-by-stage prompts – common mistakes to avoid free. no catch. happy new year. like + reply "2026" for the complete workflow pack. (must be following for DM)

KV Cache Explosion: During inference, the system caches the key-value pairs to avoid recomputation. But this cache grows linearly with context length per layer, per head. For a large model with 80+ layers and dozens of attention heads, a 1M token context means hundreds of gigabytes just for the cache.

What is the technical reason that Opus 4.5's context window is the size of a tiny walnut?