X / Twitter
OpenAI VP Science Kevin Weil
Weil shared a remarkable case: a user named Paul combined ChatGPT and AlphaFold to design a personalized mRNA vaccine plan for his dog with cancer, successfully bringing the condition under control. Weil sees this as the future of personalized medicine and called on the industry to accelerate the transfer of these capabilities from veterinary to human medicine.
https://x.com/kevinweil/status/2037420429430919674
OpenAI CEO Sam Altman
Altman also took a strong interest in Paul’s dog’s mRNA vaccine case and met Paul personally, quoting him: “AI gave me, an ordinary person, the capabilities of a research institution.” That same week, Altman announced an important Stargate milestone: the first steel beams had been erected at the Michigan data center campus developed with Oracle and Related Digital, accelerating physical infrastructure delivery.
https://x.com/sama/status/2037396826060673188
https://x.com/sama/status/2037610000122839116
Box CEO Aaron Levie
Levie announced Box’s native plugin in Codex, letting users connect any content files in Box directly to coding agents and automate file workflows. He demonstrated using Codex to process financial reports and extract structured data, aiming to connect content management more closely with AI agents and reduce manual work.
https://x.com/levie/status/2037383187245273477
Google Labs VP Josh Woodward
Woodward offered three updates: a practical prompting technique for visual content creation; Gemini’s new ability to import memories and conversation histories from other chatbots through Settings > Import memory; and promotion of video generation with Veo in Gemini at Veo 3’s release, encouraging more users to try creating videos.
https://x.com/joshwoodward/status/2037332341270733259
https://x.com/joshwoodward/status/2037674926715605331
Vercel CEO Guillermo Rauch
Rauch posted a brief but widely resonant joke: “1961: We should ship a CLI / 2026: We should ship a CLI.” The comparison humorously highlights how some software engineering needs never change. It received more than 345 likes, reflecting developers’ recognition of command-line tools’ enduring appeal.
https://x.com/rauchg/status/2037384491677933811
Every CEO Dan Shipper
Shipper held Every Demo Day this week, attracting more than 1200 participants. He also published an article on compound engineering, advocating an engineering philosophy truly designed for the AI era, and launched a plugin he said could change how teams work.
https://x.com/danshipper/status/2037589016371974486
Swyx (AI Dot Engineer)
Swyx shared an everyday productivity example: feeding years of Mac configuration blog posts directly into Claude, having AI turn personal technology-stack preferences into a bash script in one step, then publishing it as a reusable Skill. He also discovered create-context-graph, which builds key entity relationship graphs for 22 major industry domains with a single command. He believes it can serve as a “social graph layer” on top of any application. The post received more than 380 likes.
https://x.com/swyx/status/2037370812123209943
https://x.com/swyx/status/2037620876179537989
South Park Commons Partner Aditya Agarwal
Agarwal offered a distinctive view of AI startup strategy: most founders try hard to avoid competing directly with underlying models, but he encourages the opposite approach—actively betting on leaps in model capabilities 12 to 18 months from now. What models cannot do well today represents tomorrow’s most valuable differentiation opportunities. He also stressed that “procrastination is deadly,” urging founders to act decisively during the AI wave.
https://x.com/adityaag/status/2037531035978874936
https://x.com/adityaag/status/2037562946423767168
Peter Yang (Roblox product)
Yang shared how to build an /exec-review AI Skill that reproduces executives’ decision-making frameworks to obtain leadership-level product feedback at any time, with a real example from a Meta VP. He also mentioned a stronger “Capybara” model series that is larger and smarter than Opus, prompting reflections on progress toward AGI.
https://x.com/petergyang/status/2037543771580289154
Builder Zara Zhang
Zhang made a pointed observation: the real bottleneck in AI products is not AI’s context window but humans’ own context window. She further argued that the central challenge for the next generation of AI products is reconciling the tension between “human fallibility” and “agents’ near-infallibility,” posing a new question for product design.