探索

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

RT @anduriltech: "I started Anduril because I wanted to get people out of the tech industry, working on problems that I thought were not so…

RT @anduriltech: "I started Anduril because I wanted to get people out of the tech industry, working on problems that I thought were not so…

Co-Founder, American Dynamism. General Partner @a16z. Catholic. Mother. American. 🇺🇸 🚀💪

avatar for Katherine Boyle
Katherine Boyle
Opus 4.5 is close enough to my skill level that watching it work through a bug (mistakes and all) feels eerily like a brain upload.

Opus 4.5 is close enough to my skill level that watching it work through a bug (mistakes and all) feels eerily like a brain upload.

Opening portals to VR without headsets at @portalvr_io. Problems soluble, potential to improve invariant.

avatar for gfodor.id
gfodor.id
Windows 11 通知中心将重新带回 Windows 10 的日历日程功能，但没想到是通过 WebView 2 加载而非系统原生的。当用户点击通知中心的日程时任务管理器会加载多个 WebView 2 进程，并且 CPU 占用率也快速提升，使用 WebView 2 开发可以减少微软的开发成本，只不过体验上与原生版没法比。

Windows 11 通知中心将重新带回 Windows 10 的日历日程功能，但没想到是通过 WebView 2 加载而非系统原生的。当用户点击通知中心的日程时任务管理器会加载多个 WebView 2 进程，并且 CPU 占用率也快速提升，使用 WebView 2 开发可以减少微软的开发成本，只不过体验上与原生版没法比。

查看全文：https://t.co/XCAuOyvQZf

avatar for 蓝点网
蓝点网
RT @IdeasTesla: This is phenomenal. Best example of American Dynamism today, deeply inspiring. Awesome work @RadiantNuclear team, @DougBern…

RT @IdeasTesla: This is phenomenal. Best example of American Dynamism today, deeply inspiring. Awesome work @RadiantNuclear team, @DougBern…

GP @a16z — Building American Dynamism 🇺🇸 — Anthropologist — Formerly Founder/CEO @OpenDNS — Lokah Samastah Sukhino Bhavantu

avatar for David Ulevitch 🇺🇸
David Ulevitch 🇺🇸
OK, status update. Starting from absolutely nothing ~5 hours ago except a big ol' plan document, I turned that into over 350 beads (we got a bunch of new testing beads), I now have conjured up ~11k lines of code, about 8k of which is the core code and the rest is testing code (see screenshot). Around 204 commits so far. Probably at least 25 agents have been involved at some point or other. If you want to look at the actual Agent Mail messages, I used the handy export feature to publish this as a static website you can see here: https://t.co/QKP9dCfwTq So how far did we get already? You can see Claude's full assessment in the attached screenshot. Here's its bottom line conclusion, though: Key Insights 1. The product is USABLE NOW - 151 tests pass, binary compiles, all core commands work. The closed rate understates this because open issues are heavily weighted toward testing. 2. Testing is the biggest gap - ~40% of remaining work. This is actually good news - it means implementation is largely done and what remains is verification. 3. High velocity - 2.9 hour average lead time shows issues get completed, not stalled. 4. Phases 2-4 are future work - Advanced features (local semantic search, decision logging, starter playbooks) are explicitly deferred. 5. The 14 open epics are misleading - Most are testing-focused sub-epics or future phases, not blockers. Bottom Line For a "can I use this tool effectively" definition: ~85-90% done. The core ACE pipeline (Generate context → Reflect on sessions → Curate playbook → Validate scientifically) is complete and functional. What remains is mostly test coverage, polish, and future-phase features. If this were a startup product, you'd say: "MVP shipped, now hardening for production."

OK, status update. Starting from absolutely nothing ~5 hours ago except a big ol' plan document, I turned that into over 350 beads (we got a bunch of new testing beads), I now have conjured up ~11k lines of code, about 8k of which is the core code and the rest is testing code (see screenshot). Around 204 commits so far. Probably at least 25 agents have been involved at some point or other. If you want to look at the actual Agent Mail messages, I used the handy export feature to publish this as a static website you can see here: https://t.co/QKP9dCfwTq So how far did we get already? You can see Claude's full assessment in the attached screenshot. Here's its bottom line conclusion, though: Key Insights 1. The product is USABLE NOW - 151 tests pass, binary compiles, all core commands work. The closed rate understates this because open issues are heavily weighted toward testing. 2. Testing is the biggest gap - ~40% of remaining work. This is actually good news - it means implementation is largely done and what remains is verification. 3. High velocity - 2.9 hour average lead time shows issues get completed, not stalled. 4. Phases 2-4 are future work - Advanced features (local semantic search, decision logging, starter playbooks) are explicitly deferred. 5. The 14 open epics are misleading - Most are testing-focused sub-epics or future phases, not blockers. Bottom Line For a "can I use this tool effectively" definition: ~85-90% done. The core ACE pipeline (Generate context → Reflect on sessions → Curate playbook → Validate scientifically) is complete and functional. What remains is mostly test coverage, polish, and future-phase features. If this were a startup product, you'd say: "MVP shipped, now hardening for production."

Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM

avatar for Jeffrey Emanuel
Jeffrey Emanuel
我现在每天都在高频用 Gemini，ChatGPT 用的频率低了，一周也没几次。 Google One 的套餐只要 $19.99/月，可以开家庭组 6 个人共享。包含 Gemini Pro，以及 Drive、Gmail、Photos 共用的 2T 存储容量，还有 Antigravity 😂 这个定价，长远来看，其他 AI 怎么和谷歌打

我现在每天都在高频用 Gemini，ChatGPT 用的频率低了，一周也没几次。 Google One 的套餐只要 $19.99/月，可以开家庭组 6 个人共享。包含 Gemini Pro，以及 Drive、Gmail、Photos 共用的 2T 存储容量，还有 Antigravity 😂 这个定价，长远来看，其他 AI 怎么和谷歌打

程序员，开发过 2 个 iOS App，几个不知名的浏览器插件，几个小程序。喜欢看动画、漫画、轻小说和网文。希望早日成为自宅警備員。

avatar for Plusye
Plusye